技术领域technical field
本发明涉及一种基于模板匹配的MELF元件定位与检测方法。The invention relates to a method for locating and detecting MELF elements based on template matching.
背景技术Background technique
机器视觉在表面贴装技术(SMT)的应用已经越来越成熟，在贴装过程中，元件的精确定位和检测对整条SMT生产线的效率有着重要是影响。The application of machine vision in surface mount technology (SMT) has become more and more mature. During the placement process, the precise positioning and detection of components has an important impact on the efficiency of the entire SMT production line.
MELF是一种圆柱体的封装形式，两端有金属帽电极，通常有晶圆电阻、贴式电感、贴式二极管。目前已有的检测方法，主要是针对片式元件，球形引脚元件和矩形引脚元件，很少有针对圆柱体元件的研究。在特定的光照条件下,MELF元件的图像几何外形特征均表现为比较规则的矩形区域,检测算法的目标为从获取的图像中提取一个可以描述元件位姿的矩形，但在实际应用中，由于受到光源控制器制作工艺的限制，采集到的图像中元件边缘有较大的变化，且其圆柱形表面使得在接受正向光照时会出现元件表面灰度值分布不均匀的情况，给后续图像分割提取元件轮廓带来很大的困难，除此，传统的模板匹配法不适合对带旋转角度的元件进行检测，计算量大，执行速度慢，消耗时间长，难以满足贴装速度的要求。MELF is a cylindrical packaging form with metal cap electrodes at both ends, usually wafer resistors, bonded inductors, and bonded diodes. The existing detection methods are mainly for chip components, spherical lead components and rectangular lead components, and there are few researches for cylindrical components. Under specific lighting conditions, the image geometry features of MELF components are relatively regular rectangular areas. The goal of the detection algorithm is to extract a rectangle that can describe the component pose from the acquired image. However, in practical applications, due to Limited by the manufacturing process of the light source controller, the edge of the component in the collected image has a large change, and its cylindrical surface makes the distribution of gray value on the surface of the component uneven when it receives positive light, which will affect the follow-up image. Segmenting and extracting component outlines brings great difficulties. In addition, the traditional template matching method is not suitable for detecting components with rotation angles. It has a large amount of calculation, slow execution speed, and takes a long time to meet the requirements of placement speed.
发明内容Contents of the invention
本发明的目的是为了解决传统模板匹配算法对带旋转角度的元件进行检测时存在计算量大、执行速度慢，导致元件定位与检测速度慢的问题，而提出一种基于模板匹配的MELF元件定位与检测方法。The purpose of the present invention is to solve the problem that the traditional template matching algorithm has a large amount of calculation and slow execution speed when detecting components with rotation angles, resulting in slow component positioning and detection speed, and proposes a MELF component positioning based on template matching and detection methods.
一种基于模板匹配的MELF元件定位与检测方法，所述MELF元件定位与检测方法通过以下步骤实现：A MELF element location and detection method based on template matching, the MELF element location and detection method is realized by the following steps:
步骤一、采用光学拍摄系统获取MELF元件的原始MELF元件图像；Step 1. Obtain the original MELF component image of the MELF component by using an optical shooting system;
步骤二、选用固定阈值对步骤一获得的原始MELF元件图像进行阈值分割，得到二值化预处理后的图像，并计算二值化预处理后的图像中非零像素点的个数；Step 2, select a fixed threshold to carry out threshold segmentation on the original MELF component image obtained in step 1, obtain the image after binarization preprocessing, and calculate the number of non-zero pixels in the image after binarization preprocessing;
步骤三、判断步骤二得到的非零像素点的个数是否达到原始MELF元件图像像素总数的相应倍数，若否，则结束MELF元件检测过程，返回相应的错误码；若是，继续执行步骤四；Step 3, judging whether the number of non-zero pixel points obtained in step 2 reaches the corresponding multiple of the total number of pixels in the original MELF component image, if not, then end the MELF component detection process, and return a corresponding error code; if so, continue to perform step 4;
步骤四、根据输入的MELF元件的长度和宽度信息建立旋转角度为0°的模板图像，对旋转角度为0°的模板图像以1度为步长进行旋转，得到包括旋转角度为0°的模板图像在内的旋转角度在[-30°,30°]之间的所有模板图像；Step 4: Establish a template image with a rotation angle of 0° according to the length and width information of the input MELF component, rotate the template image with a rotation angle of 0° at a step of 1 degree, and obtain a template including a rotation angle of 0° All template images whose rotation angle is between [-30°, 30°];
步骤五、采用图像高斯金字塔计算方法，将步骤一获得的原始MELF元件图像缩小至四分之一作为缩小后的MELF元件图像，将步骤四获得的模板图像缩小至四分之一作为缩小后的模板图像；其中，缩小后的模板图像包括缩小后的旋转角度为0°的模板图像和缩小后的旋转角度在[-30°,30°]之间的模板图像；Step 5. Using the image Gaussian pyramid calculation method, the original MELF component image obtained in step 1 is reduced to 1/4 as the reduced MELF component image, and the template image obtained in step 4 is reduced to 1/4 as the reduced image A template image; wherein, the reduced template image includes a template image whose reduced rotation angle is 0° and a template image whose reduced rotation angle is between [-30°, 30°];
步骤六、分别对步骤五获得的缩小后的MELF元件图像以及步骤一获得的原始MELF元件图像进行Canny边缘检测得到边缘图像，之后分别对获得的边缘图像进行非操作，之后分别计算非操作后的边缘图像中非零像素点到最近零像素点的距离，作为缩小后的MELF元件图像的距离变换图像和原始MELF元件图像的距离变换图像；Step 6. Perform Canny edge detection on the reduced MELF component image obtained in step 5 and the original MELF component image obtained in step 1 to obtain an edge image, then perform non-operation on the obtained edge image, and then calculate the non-operated The distance from the non-zero pixel point to the nearest zero pixel point in the edge image is used as the distance transformed image of the reduced MELF component image and the distance transformed image of the original MELF component image;
步骤七、用步骤四获得的模板图像搜索待匹配图像，待匹配图像上被模板图像覆盖的区域作为子区域图像，采用模板匹配计算公式：对子区域图像与模板图像的相似性与差异性进行模板匹配计算，获取最佳匹配模板和最佳匹配模板在原始MELF元件图像中的最佳匹配位置；其中，T(m,n)表示模板图像T在(m,n)处的灰度值；s(i,j)(m,n)表示与子区域图像s(i,j)中对应的(m,n)处的灰度值；Step 7. Use the template image obtained in step 4 to search for the image to be matched, and the area covered by the template image on the image to be matched is used as a sub-region image, and the template matching calculation formula is used: Perform template matching calculation on the similarity and difference between the sub-region image and the template image, and obtain the best matching template and the best matching position of the best matching template in the original MELF component image; where T(m,n) represents the template The gray value of image T at (m, n); s(i, j) (m, n) represents the gray value at (m, n) corresponding to the sub-region image s(i, j) ;
步骤八、对步骤一获取的原始MELF元件图像采用Canny边缘检测得到带干扰点的MELF边缘图像，利用步骤七得到的最佳匹配模板和最佳匹配模板在原始MELF元件图像中的最佳匹配位置的匹配结果，根据图像边缘点的相关性从带干扰点的MELF边缘图像中提取出关键边缘点，以确定原始MELF图像中MELF元件的整体轮廓，完成MELF元件的定位过程；Step 8. Use Canny edge detection on the original MELF component image obtained in step 1 to obtain a MELF edge image with interference points, and use the best matching template obtained in step 7 and the best matching position of the best matching template in the original MELF component image According to the matching results of the image edge points, the key edge points are extracted from the MELF edge image with interference points according to the correlation of the image edge points, so as to determine the overall outline of the MELF element in the original MELF image, and complete the positioning process of the MELF element;
步骤九、利用步骤八获得的所有关键边缘点形成最小外接矩形，用最小外接矩形表示MELF元件在原始MELF元件图像中的位置，最小外接矩形的中心坐标表示MELF元件的中心坐标，最小外接矩形的旋转角度表示MELF元件的旋转角度；在最小外接矩形的中心坐标和旋转角度不变的情况下，将最小外接矩形的长度和宽度放大1.1倍之后形成向外增加偏置量后的矩形；再在最小外接矩形的中心坐标和旋转角度的情况下，将最小外接矩形的长度和宽度缩小0.9倍之后形成向内增加偏置量后的矩形；Step 9: Use all key edge points obtained in step 8 to form a minimum circumscribed rectangle, use the minimum circumscribed rectangle to represent the position of the MELF component in the original MELF component image, the center coordinates of the minimum circumscribed rectangle represent the center coordinates of the MELF component, and the minimum circumscribed rectangle The rotation angle represents the rotation angle of the MELF element; in the case that the center coordinates and rotation angle of the minimum circumscribed rectangle remain unchanged, the length and width of the minimum circumscribed rectangle are enlarged by 1.1 times to form a rectangle with an outwardly increased offset; and then In the case of the center coordinates and rotation angle of the minimum circumscribed rectangle, the length and width of the minimum circumscribed rectangle are reduced by 0.9 times to form a rectangle with an inwardly increased offset;
步骤十、根据步骤九获得的向外增加偏置量后的矩形或向内增加偏置量后的矩形内非零像素的个数，判断检测得到的MELF元件位置是否正确，若不正确，则结束MELF元件的检测过程，返回相应的错误码，若正确，则执行步骤十一；Step 10. According to the number of non-zero pixels in the rectangle after increasing the offset amount outward or the rectangle after increasing the offset amount inward obtained in step 9, judge whether the position of the detected MELF element is correct, if not, then End the detection process of the MELF component, return the corresponding error code, if it is correct, go to step 11;
步骤十一、根据最小外接矩形的尺寸判断元件的长度和宽度是否在容差范围内，若是，则结束MELF元件的定位与检测过程并输出MELF元件的检测结果、中心坐标和旋转角度，若否，则结束MELF元件的检测过程，返回相应的错误码。Step 11. Determine whether the length and width of the element are within the tolerance range according to the size of the smallest circumscribed rectangle, if yes, end the positioning and detection process of the MELF element and output the detection result, center coordinates and rotation angle of the MELF element, if not , then the detection process of the MELF component ends, and the corresponding error code is returned.
本发明的有益效果为：The beneficial effects of the present invention are:
本发明方法解决了由于MELF元件的圆柱体形状特性导致的在接受正向光照时采集到的图像表面灰度不均匀，不利于后续图像分割实现元件定位的问题。通过Canny边缘检测得到边缘图像，之后对边缘图像进行非操作，之后计算非操作后的边缘图像中非零像素点到最近零像素点的距离，得到缩小后的MELF元件图像的距离变换图像和原始MELF元件图像的距离变换图像，利用MELF元件的距离变换图像并采用模板匹配方法，获取最终精确匹配的最佳匹配模板图像和其在原始MELF图像中的最佳匹配位置；在带干扰点的MELF边缘图像中利用图像边缘的相关性提取关键边缘点，用包含所有关键边缘点的最小外接矩形表示MELF元件的整体轮廓进行模板匹配的技术方案，避免了元件边缘变化影响检测过程，将检测精度保持在97％左右；之后采用旋转角度在[-30°,30°]之间的模板图像的模板图像，来检测带有旋转角度的待匹配元件；利用高斯金字塔计算方法对原始MELF元件图像和旋转角度在[-30°,30°]之间的模板图像的模板图像进行图像数据的压缩，得到缩小后的MELF元件图像和缩小后的模板图像图像，能够减少模板匹配计算过程中搜索位置的个数和每次匹配的计算量，具有提高模板匹配计算效率的好处，与现有模板匹配计算相比过程相比，将模板匹配计算量降低50％左右和将计算耗时减少60％左右；The method of the invention solves the problem that the gray scale of the surface of the image collected when receiving forward light is not uniform due to the cylinder shape characteristic of the MELF element, which is not conducive to the subsequent image segmentation to realize element positioning. The edge image is obtained by Canny edge detection, and then the edge image is not operated, and then the distance from the non-zero pixel point to the nearest zero pixel point in the non-operated edge image is calculated to obtain the distance transformed image of the reduced MELF component image and the original The distance transformed image of the MELF component image uses the distance transformed image of the MELF component and adopts the template matching method to obtain the best matching template image and its best matching position in the original MELF image; In the edge image, the correlation of image edges is used to extract key edge points, and the minimum circumscribed rectangle containing all key edge points is used to represent the overall outline of MELF components for template matching, which avoids the influence of component edge changes on the detection process and maintains the detection accuracy. At about 97%; then use the template image of the template image with the rotation angle between [-30°, 30°] to detect the component to be matched with the rotation angle; use the Gaussian pyramid calculation method to calculate the original MELF component image and rotation The template image of the template image whose angle is between [-30°, 30°] is compressed to obtain the reduced MELF component image and the reduced template image image, which can reduce the number of search positions in the template matching calculation process. Compared with the existing template matching calculation process, the template matching calculation amount is reduced by about 50% and the calculation time consumption is reduced by about 60%;
本发明方法的鲁棒性好，当模板大小发生变化时，仍能进行匹配，并且对匹配成功与否进行检测，将MELF元件的定位与检测的正确率提高至95-98％。The method of the invention has good robustness, can still perform matching when the size of the template changes, and detects whether the matching is successful, and improves the correct rate of positioning and detection of the MELF element to 95-98%.
附图说明Description of drawings
图1为本发明的流程图；Fig. 1 is a flow chart of the present invention;
图2为本发明涉及的光学拍摄系统获取的MELF元件的原始MELF元件图像；Fig. 2 is the original MELF component image of the MELF component that the optical photographing system that the present invention obtains;
图3为本发明涉及的对原始MELF元件图像进行Canny边缘检测得到边缘图像；Fig. 3 is carried out Canny edge detection to the original MELF element image that the present invention relates to and obtains the edge image;
图4为本发明涉及的原始MELF元件图像的距离变换图像；Fig. 4 is the distance transformation image of the original MELF component image that the present invention relates to;
图5为本发明涉及的旋转角度为0°的模板图像；Fig. 5 is the template image that the rotation angle involved in the present invention is 0 °;
图6为本发明涉及的缩小后的MELF元件图像的距离变换图像；Fig. 6 is the distance transformation image of the reduced MELF element image involved in the present invention;
图7为本发明涉及的缩小后的模板图像；Fig. 7 is the reduced template image involved in the present invention;
图8为本发明涉及的初次粗略匹配结果示意图，其中，初次粗略匹配的最佳匹配位置作为矩形的左上顶点；Fig. 8 is a schematic diagram of the initial rough matching result involved in the present invention, wherein the best matching position of the initial rough matching is taken as the upper left vertex of the rectangle;
图9为本发明涉及的在缩小后的MELF元件图像的距离变换图像中截取的感兴趣区域ROI；Fig. 9 is the region of interest ROI intercepted in the distance transformed image of the reduced MELF element image involved in the present invention;
图10为本发明涉及的第二次粗略匹配结果示意图，其中，矩形的左上顶点集合为最佳匹配位置候选集；Fig. 10 is a schematic diagram of the second rough matching result involved in the present invention, wherein the upper left vertex set of the rectangle is the best matching position candidate set;
图11为本发明涉及的第二次粗略匹配的最佳匹配模板；Fig. 11 is the best matching template of the second rough matching involved in the present invention;
图12为本发明涉及的第二次粗略匹配的最佳匹配位置截取的感兴趣区域ROI；Fig. 12 is the region of interest ROI intercepted at the best matching position of the second rough matching involved in the present invention;
图13为本发明涉及的原始MELF图像中精确匹配的最佳匹配模板图像；Fig. 13 is the best matching template image accurately matched in the original MELF image involved in the present invention;
图14为本发明涉及的原始MELF图像精确匹配结果示意图，其中，矩形的左上顶点为原始MELF元件图像中的最佳匹配位置；Fig. 14 is a schematic diagram of the exact matching result of the original MELF image involved in the present invention, wherein the upper left vertex of the rectangle is the best matching position in the original MELF component image;
图15为本发明涉及的MELF元件的定位图像，矩形框为包含所有关键边缘点的最小外接矩形，代表元件在原始MELF图像中的位置。Fig. 15 is a positioning image of MELF components involved in the present invention, and the rectangular frame is the smallest circumscribed rectangle containing all key edge points, representing the position of the component in the original MELF image.
具体实施方式Detailed ways
具体实施方式一：Specific implementation mode one:
本实施方式的一种基于模板匹配的MELF元件定位与检测方法，如图1所示，所述MELF元件定位与检测方法通过以下步骤实现：A kind of MELF element localization and detection method based on template matching of the present embodiment, as shown in Figure 1, described MELF element localization and detection method are realized through the following steps:
步骤一、采用光学拍摄系统获取MELF元件的原始MELF元件图像；Step 1. Obtain the original MELF component image of the MELF component by using an optical shooting system;
步骤二、选用固定阈值对步骤一获得的原始MELF元件图像进行阈值分割，得到二值化预处理后的图像，并计算二值化预处理后的图像中非零像素点的个数；Step 2, select a fixed threshold to carry out threshold segmentation on the original MELF component image obtained in step 1, obtain the image after binarization preprocessing, and calculate the number of non-zero pixels in the image after binarization preprocessing;
步骤三、判断步骤二得到的非零像素点的个数是否达到原始MELF元件图像像素总数的相应倍数，若否，则结束MELF元件检测过程，返回相应的错误码；若是，继续执行步骤四；Step 3, judging whether the number of non-zero pixel points obtained in step 2 reaches the corresponding multiple of the total number of pixels in the original MELF component image, if not, then end the MELF component detection process, and return a corresponding error code; if so, continue to perform step 4;
步骤四、根据输入的MELF元件的长度和宽度信息建立旋转角度为0°的模板图像，对旋转角度为0°的模板图像以1度为步长进行旋转，得到包括旋转角度为0°的模板图像在内的旋转角度在[-30°,30°]之间的所有模板图像；Step 4: Establish a template image with a rotation angle of 0° according to the length and width information of the input MELF component, rotate the template image with a rotation angle of 0° at a step of 1 degree, and obtain a template including a rotation angle of 0° All template images whose rotation angle is between [-30°, 30°];
步骤五、采用图像高斯金字塔计算方法，将步骤一获得的原始MELF元件图像缩小至四分之一作为缩小后的MELF元件图像，将步骤四获得的模板图像缩小至四分之一作为缩小后的模板图像；其中，缩小后的模板图像包括缩小后的旋转角度为0°的模板图像和缩小后的旋转角度在[-30°,30°]之间的模板图像；这种对原始MELF元件图像和带有角度的模板图像进行的图像数据压缩，能够减少步骤七模板匹配计算过程中搜索位置的个数和每次匹配的计算量，具有提高模板匹配计算效率的好处。Step 5. Using the image Gaussian pyramid calculation method, the original MELF component image obtained in step 1 is reduced to 1/4 as the reduced MELF component image, and the template image obtained in step 4 is reduced to 1/4 as the reduced image Template image; Wherein, the template image after reduction comprises the template image that the rotation angle after reduction is 0 ° and the template image that the rotation angle after reduction is between [-30 °, 30 °]; This is to original MELF component image The image data compression performed with the template image with an angle can reduce the number of search positions and the calculation amount of each match in the template matching calculation process of step 7, and has the advantage of improving the template matching calculation efficiency.
步骤六、分别对步骤五获得的缩小后的MELF元件图像以及步骤一获得的原始MELF元件图像进行Canny边缘检测得到边缘图像，之后分别对获得的边缘图像进行非操作，之后分别计算非操作后的边缘图像中非零像素点到最近零像素点的距离，作为缩小后的MELF元件图像的距离变换图像和原始MELF元件图像的距离变换图像；Step 6. Perform Canny edge detection on the reduced MELF component image obtained in step 5 and the original MELF component image obtained in step 1 to obtain an edge image, then perform non-operation on the obtained edge image, and then calculate the non-operated The distance from the non-zero pixel point to the nearest zero pixel point in the edge image is used as the distance transformed image of the reduced MELF component image and the distance transformed image of the original MELF component image;
步骤七、用步骤四获得的模板图像搜索待匹配图像，待匹配图像上被模板图像覆盖的区域作为子区域图像，采用模板匹配计算公式：对子区域图像的像素与模板图像的对应各点灰度差平方的和来表示的子区域图像与模板图像的相似性与差异性进行模板匹配计算，获取最佳匹配模板和最佳匹配模板在原始MELF元件图像中的最佳匹配位置；其中，T(m,n)表示模板图像T在(m,n)处的灰度值；s(i,j)(m,n)表示与子区域图像s(i,j)中对应的(m,n)处的灰度值；Step 7. Use the template image obtained in step 4 to search for the image to be matched, and the area covered by the template image on the image to be matched is used as a sub-region image, and the template matching calculation formula is used: The template matching calculation is performed on the similarity and difference between the sub-region image and the template image represented by the sum of the squares of the gray level differences between the pixels of the sub-region image and the corresponding points of the template image, and the best matching template and the best matching template are obtained. The best matching position in the original MELF component image; among them, T(m,n) represents the gray value of the template image T at (m,n); s(i,j) (m,n) represents the sub-region The gray value at the corresponding (m, n) in the image s(i, j) ;
步骤八、对步骤一获取的原始MELF元件图像采用Canny边缘检测得到带干扰点的MELF边缘图像，利用步骤七得到的最佳匹配模板和最佳匹配模板在原始MELF元件图像中的最佳匹配位置的匹配结果，根据图像边缘点的相关性从带干扰点的MELF边缘图像中提取出关键边缘点，以确定原始MELF图像中MELF元件的整体轮廓，完成MELF元件的定位过程；Step 8. Use Canny edge detection on the original MELF component image obtained in step 1 to obtain a MELF edge image with interference points, and use the best matching template obtained in step 7 and the best matching position of the best matching template in the original MELF component image According to the matching results of the image edge points, the key edge points are extracted from the MELF edge image with interference points according to the correlation of the image edge points, so as to determine the overall outline of the MELF element in the original MELF image, and complete the positioning process of the MELF element;
步骤九、利用步骤八获得的所有关键边缘点形成最小外接矩形，用最小外接矩形表示MELF元件在原始MELF元件图像中的位置，最小外接矩形的中心坐标表示MELF元件的中心坐标，最小外接矩形的旋转角度表示MELF元件的旋转角度；在最小外接矩形的中心坐标和旋转角度不变的情况下，将最小外接矩形的长度和宽度放大1.1倍之后形成向外增加偏置量后的矩形；再在最小外接矩形的中心坐标和旋转角度的情况下，将最小外接矩形的长度和宽度缩小0.9倍之后形成向内增加偏置量后的矩形；Step 9: Use all key edge points obtained in step 8 to form a minimum circumscribed rectangle, use the minimum circumscribed rectangle to represent the position of the MELF component in the original MELF component image, the center coordinates of the minimum circumscribed rectangle represent the center coordinates of the MELF component, and the minimum circumscribed rectangle The rotation angle represents the rotation angle of the MELF element; in the case that the center coordinates and rotation angle of the minimum circumscribed rectangle remain unchanged, the length and width of the minimum circumscribed rectangle are enlarged by 1.1 times to form a rectangle with an outwardly increased offset; and then In the case of the center coordinates and rotation angle of the minimum circumscribed rectangle, the length and width of the minimum circumscribed rectangle are reduced by 0.9 times to form a rectangle with an inwardly increased offset;
步骤十、根据步骤九获得的向外增加偏置量后的矩形或向内增加偏置量后的矩形内非零像素的个数，判断检测得到的MELF元件位置是否正确，若不正确，则结束MELF元件的检测过程，返回相应的错误码，若正确，则执行步骤十一；Step 10. According to the number of non-zero pixels in the rectangle after increasing the offset amount outward or the rectangle after increasing the offset amount inward obtained in step 9, judge whether the position of the detected MELF element is correct, if not, then End the detection process of the MELF component, return the corresponding error code, if it is correct, go to step 11;
步骤十一、根据最小外接矩形的尺寸判断元件的长度和宽度是否在容差范围内，若是，则结束MELF元件的定位与检测过程并输出MELF元件的检测结果、中心坐标和旋转角度，若否，则结束MELF元件的检测过程，返回相应的错误码。Step 11. Determine whether the length and width of the element are within the tolerance range according to the size of the smallest circumscribed rectangle, if yes, end the positioning and detection process of the MELF element and output the detection result, center coordinates and rotation angle of the MELF element, if not , then the detection process of the MELF component ends, and the corresponding error code is returned.
具体实施方式二：Specific implementation mode two:
与具体实施方式一不同的是，本实施方式的基于模板匹配的MELF元件定位与检测方法，步骤四所述根据输入的MELF元件的长度和宽度信息建立旋转角度为0°的模板图像，对旋转角度为0°的模板图像以1度为步长进行旋转，得到旋转角度在[-30°,30°]之间的所有模板图像的过程为，Different from Embodiment 1, in the MELF component positioning and detection method based on template matching of the present embodiment, the template image with a rotation angle of 0° is established according to the length and width information of the input MELF component described in step 4, and the rotation angle is 0 °. The template image with an angle of 0° is rotated with a step size of 1 degree, and the process of obtaining all template images with a rotation angle between [-30°, 30°] is,
步骤四一、根据输入MELF元件的长度和宽度信息获取旋转角度为0°的模板信息，建立旋转角度为0°的模板图像；其中，MELF元件的长度对应旋转角度为0°模板图像的宽度，MELF元件的宽度对应旋转角度为0°模板图像的高度；Step 41, according to the length and width information of the input MELF element, obtain the template information with a rotation angle of 0°, and establish a template image with a rotation angle of 0°; wherein, the length of the MELF element corresponds to the width of the template image with a rotation angle of 0°, The width of the MELF element corresponds to the height of the template image whose rotation angle is 0°;
步骤四二、对旋转角度为0°的模板图像以1度为步长进行旋转，并通过w'＝h|sinθ|+w|cosθ|，h'＝w|sinθ|+h|cosθ|计算旋转角度在[-30°,30°]之间的模板信息，建立旋转角度在[-30°,30°]之间的模板图像；其中，θ表示旋转角度，w、h分别表示旋转角度为0°的模板图像的宽度和高度，w'、h'为旋转后的模板图像的宽度和高度，用于步骤七的模板匹配。Step 42: Rotate the template image with a rotation angle of 0° with a step size of 1 degree, and calculate by w'=h|sinθ|+w|cosθ|, h'=w|sinθ|+h|cosθ| The template information with the rotation angle between [-30°, 30°], establishes the template image with the rotation angle between [-30°, 30°]; where θ represents the rotation angle, and w and h respectively represent the rotation angle of 0° is the width and height of the template image, w' and h' are the width and height of the rotated template image, which are used for template matching in step 7.
具体实施方式三：Specific implementation mode three:
与具体实施方式一或二不同的是，本实施方式的基于模板匹配的MELF元件定位与检测方法，步骤五所述图像高斯金字塔计算方法的具体计算过程为，Different from the specific embodiment 1 or 2, the specific calculation process of the image Gaussian pyramid calculation method described in step 5 of the MELF element positioning and detection method based on template matching in this embodiment is as follows:
步骤五一、设要获得的金字塔图像包括i+1层图像，金字塔图像中每一层图像都由同一张原始MELF元件图像获得，金字塔图像的层级采用由下至上的次序编号，且层级越高图像越小，最高的第i+1层代表金字塔图像的最高层级，Gi表示层级i的金字塔图像；则将Gi与高斯内核
步骤五二、设原始图像G0表示输入的原始MELF元件图像或者模板图像，按照步骤五一所述Gi+1的采样计算过程，对输入的原始图像G0进行迭代计算，直到得到的图像为原始图像G0的四分之一，即得到缩小后的MELF元件图像或者缩小后的模板图像。Step 52. Let the original image G0 represent the input original MELF component image or template image, and perform iterative calculation on the input original image G0 according to the sampling calculation process of Gi+1 described in step 51 until the obtained image is a quarter of the original image G0 , that is, the reduced MELF component image or the reduced template image is obtained.
具体实施方式四：Specific implementation mode four:
与具体实施方式三不同的是，本实施方式的基于模板匹配的MELF元件定位与检测方法，步骤七所述待匹配图像中的子区域图像和模板图像之间的匹配性由相关度函数R(i,j)进行测定，相关度函数R(i,j)的计算公式：
具体实施方式五：Specific implementation mode five:
与具体实施方式一、二或四不同的是，本实施方式的基于模板匹配的MELF元件定位与检测方法，步骤七所述对子区域图像与模板图像的相似性与差异性进行模板匹配计算的过程为，Different from specific embodiments 1, 2 or 4, in the MELF element positioning and detection method based on template matching in this embodiment, the template matching calculation of the similarity and difference between the sub-region image and the template image described in step 7 The process is,
步骤七一、选取缩小后的旋转角度为0°的模板图像与缩小后的MELF元件图像的距离变换图像进行模板匹配计算，搜索得到最小相关度R(i,j)值，并将最小相关度R(i,j)值对应的位置作为初次粗略匹配的最佳匹配位置；Step 71, select the template image with a reduced rotation angle of 0° and the distance transformed image of the reduced MELF component image for template matching calculation, search for the minimum correlation R (i, j) value, and use the minimum correlation The position corresponding to the R(i,j) value is used as the best matching position for the first rough match;
步骤七二、将步骤七一得到的初次粗略匹配的最佳匹配位置作为矩形的左上顶点，在缩小后的MELF元件图像的距离变换图像中截取一个与缩小后的旋转角度为0°的模板图像相同大小的矩形，再分别向四周扩展相应的偏置量，作为下一步检测图像的感兴趣区域ROI；Step 72: Use the best matching position obtained in step 71 as the upper left vertex of the rectangle, and intercept a template image with a reduced rotation angle of 0° from the distance transformed image of the reduced MELF component image Rectangles of the same size, and then expand the corresponding offsets to the surroundings respectively, as the next step to detect the region of interest ROI of the image;
步骤七三、在步骤七二获得的感兴趣区域ROI中，用所有的缩小后的旋转角度在[-30°,30°]之间的模板图像进行模板匹配计算，搜索得到第二次粗略匹配相关度值在0.9-1.1倍的最小相关度R(i,j)值内对应的所有匹配模板图像序列号及对应的匹配位置；Step 73, in the region of interest ROI obtained in step 72, use all template images with reduced rotation angles between [-30°, 30°] to perform template matching calculations, and search to obtain the second rough match All matching template image sequence numbers and corresponding matching positions corresponding to the correlation value within the minimum correlation R(i, j) value of 0.9-1.1 times;
步骤七四、将步骤七三获得的所有匹配模板图像形成的候选集中最小相关度R(i,j)值对应的模板图像作为最佳匹配模板图像，使用坐标变换得到最佳匹配模板图像的匹配位置在原始MELF元件图像的距离变换图像中对应的位置，作为矩形的左上顶点，在原始MELF元件图像的距离变换图像中截取一个与最佳匹配序列号对应的原模板图像相同大小的矩形，再分别向四周扩展相应的偏置量，作为下一步检测图像的感兴趣区域ROI；Step 74, use the template image corresponding to the minimum correlation R(i, j) value in the candidate set formed by all matching template images obtained in step 73 as the best matching template image, and use coordinate transformation to obtain the matching of the best matching template image The corresponding position in the distance transformation image of the original MELF component image is used as the upper left vertex of the rectangle, and a rectangle with the same size as the original template image corresponding to the best matching serial number is intercepted in the distance transformation image of the original MELF component image, and then Expand the corresponding offsets to the surroundings respectively, as the next step to detect the ROI of the image;
步骤七五、采用步骤七四获得的最佳匹配模板图像的最佳匹配序列号，以及最佳匹配序列号前后各三个序列号对应的模板图像，与步骤七四得到的感兴趣区域ROI和进行模板匹配计算，得到精确匹配的最佳匹配模板图像的序列号及精确匹配的最佳匹配位置，通过坐标变换得到精确匹配的最佳匹配位置在原始MELF元件图像中对应的位置。Step 75, adopt the best matching sequence number of the best matching template image obtained in step 74, and the template image corresponding to the three serial numbers before and after the best matching sequence number, and the ROI and ROI obtained in step 74 Perform template matching calculation to obtain the serial number of the best matching template image for exact matching and the best matching position for exact matching, and obtain the corresponding position of the best matching position for exact matching in the original MELF component image through coordinate transformation.
具体实施方式六：Specific implementation method six:
与具体实施方式五不同的是，本实施方式的基于模板匹配的MELF元件定位与检测方法，步骤八中所述从带干扰点的MELF边缘图像中提取出关键边缘点，以确定原始MELF图像中MELF元件的整体轮廓的过程为：Different from Embodiment 5, in the MELF component positioning and detection method based on template matching in this embodiment, the key edge points are extracted from the MELF edge image with interference points as described in step 8, so as to determine the The process of the overall outline of the MELF element is:
步骤八一、以精确匹配的最佳匹配位置在原始MELF图像中的对应位置作为矩形的左上顶点，在原始MELF元件图像的距离变换图像中截取一个与精确匹配的最佳匹配模板图像相同大小的矩形图像；Step 81: Take the corresponding position of the best matching position of exact matching in the original MELF image as the upper left vertex of the rectangle, and intercept a template image of the same size as the best matching template image of exact matching in the distance transformed image of the original MELF component image rectangle image;
步骤八二、将精确匹配的最佳匹配模板图像做归一化处理，并将精确匹配的最佳匹配模板图像的灰度值转化到[0,1]之间；将归一化后的模板图像和步骤八一截取的矩形图像进行乘操作，寻找乘操作后图像最大灰度值；Step 82: Normalize the best matching template image for exact matching, and convert the grayscale value of the best matching template image for exact matching to [0,1]; normalize the template The image and the rectangular image intercepted in step 81 are multiplied to find the maximum gray value of the image after the multiplication operation;
步骤八三、对精确匹配的最佳匹配模板图像进行非操作，再计算非零像素到最近零像素点的距离并作为模板距离变换图像；Step 83, performing a non-operation on the best matching template image of the exact match, and then calculating the distance from the non-zero pixel to the nearest zero pixel and using it as a template distance transformation image;
步骤八四、以精确匹配的最佳匹配位置在带干扰点的MELF边缘图像中对应的位置作为矩形的左上顶点，在带干扰点的MELF边缘图像中截取一个与精确匹配的最佳匹配模板图像相同大小的矩形图像，并对其采用归一化处理，将灰度值转化到[0,1]之间；Step 84, use the corresponding position of the best matching position of the exact match in the MELF edge image with the interference point as the upper left vertex of the rectangle, and intercept a best matching template image with the exact match in the MELF edge image with the interference point Rectangular images of the same size, and normalize them, convert the gray value to [0,1];
步骤八五、对步骤八三计算得到的模板距离变换图像和步骤八四归一化后的矩形图像进行乘操作，以步骤八二中得到的图像最大灰度值为阈值对乘操作后的图像进行反阈值化处理，之后和步骤八四中未归一化处理的矩形图像进行与操作，与操作后图像内的点集即为元件的关键边缘点。Step 85: Multiply the template distance transformed image calculated in step 83 and the rectangular image normalized in step 84, and use the maximum gray value of the image obtained in step 82 as a threshold value to multiply the image Perform anti-thresholding processing, and then perform an AND operation with the unnormalized rectangular image in step 84, and the point set in the image after the AND operation is the key edge point of the component.
具体实施方式七：Specific implementation mode seven:
与具体实施方式一、二、四或六不同的是，本实施方式的基于模板匹配的MELF元件定位与检测方法，步骤十所述判断检测得到的MELF元件位置是否正确的过程是，当向外增加偏置量后的矩形和向内增加偏置量后的矩形内非零像素的个数中至少有一个达到步骤二得到的非零像素点个数的0.9倍，判断检测得到的元件位置正确，否则判断出检测得到元件位置不正确。Different from specific embodiments 1, 2, 4 or 6, in the method for positioning and detecting MELF components based on template matching in this embodiment, the process of judging whether the position of the detected MELF components described in step 10 is correct is that when outward At least one of the number of non-zero pixels in the rectangle after increasing the offset and the rectangle after increasing the offset inward reaches 0.9 times the number of non-zero pixels obtained in step 2, and it is judged that the position of the detected component is correct , otherwise it is judged that the detected component position is incorrect.
实施例1：Example 1:
基于模板匹配的MELF元件定位与检测方法，所述MELF元件定位与检测方法通过以下步骤实现：Based on the template matching MELF element location and detection method, the MELF element location and detection method is realized through the following steps:
步骤一、采用光学拍摄系统获取如图2所示的MELF元件的原始MELF元件图像；Step 1, adopt the optical photographing system to obtain the original MELF component image of the MELF component as shown in Figure 2;
步骤二、选用固定阈值对步骤一获得的原始MELF元件图像进行阈值分割，得到二值化预处理后的图像，并计算二值化预处理后的图像中非零像素点的个数；Step 2, select a fixed threshold to carry out threshold segmentation on the original MELF component image obtained in step 1, obtain the image after binarization preprocessing, and calculate the number of non-zero pixels in the image after binarization preprocessing;
步骤三、判断步骤二得到的非零像素点的个数是否达到原始MELF元件图像像素总数的相应倍数，若否，则结束MELF元件检测过程，返回相应的错误码；若是，继续执行步骤四；Step 3, judging whether the number of non-zero pixel points obtained in step 2 reaches the corresponding multiple of the total number of pixels in the original MELF component image, if not, then end the MELF component detection process, and return a corresponding error code; if so, continue to perform step 4;
步骤四、根据输入的MELF元件的长度和宽度信息建立如图5所示的旋转角度为0°的模板图像，对旋转角度为0°的模板图像以1度为步长进行旋转，得到包括旋转角度为0°的模板图像在内的旋转角度在[-30°,30°]之间的所有模板图像；Step 4: Establish a template image with a rotation angle of 0° as shown in Figure 5 according to the length and width information of the input MELF element, and rotate the template image with a rotation angle of 0° at a step of 1 degree to obtain All template images with a rotation angle between [-30°, 30°] including the template image with an angle of 0°;
步骤五、采用图像高斯金字塔计算方法，将步骤一获得的原始MELF元件图像缩小至四分之一作为缩小后的MELF元件图像，将步骤四获得的模板图像缩小至四分之一作为如图7所示的缩小后的模板图像；其中，缩小后的模板图像包括缩小后的旋转角度为0°的模板图像和缩小后的旋转角度在[-30°,30°]之间的模板图像；Step 5. Use the image Gaussian pyramid calculation method to reduce the original MELF component image obtained in step 1 to 1/4 as the reduced MELF component image, and reduce the template image obtained in step 4 to 1/4 as shown in Figure 7 The reduced template image shown; wherein, the reduced template image includes a reduced rotation angle of 0 ° of the template image and a reduced rotation angle of the template image between [-30 °, 30 °];
步骤六、分别对步骤五获得的缩小后的MELF元件图像以及步骤一获得的原始MELF元件图像进行Canny边缘检测得到如图3所示的边缘图像，之后分别对获得的边缘图像进行非操作，之后分别计算非操作后的边缘图像中非零像素点到最近零像素点的距离，作为如图6所示的缩小后的MELF元件图像的距离变换图像，和如图4所示的原始MELF元件图像的距离变换图像；Step 6. Perform Canny edge detection on the reduced MELF component image obtained in step 5 and the original MELF component image obtained in step 1 to obtain the edge image shown in Figure 3, and then perform non-operation on the obtained edge image respectively, and then Calculate the distance from the non-zero pixel point to the nearest zero pixel point in the non-operated edge image respectively, as the distance transformation image of the reduced MELF component image as shown in Figure 6, and the original MELF component image as shown in Figure 4 The distance transformed image;
步骤七、用步骤四获得的模板图像搜索待匹配图像，待匹配图像上被模板图像覆盖的区域作为子区域图像，采用模板匹配计算公式：对子区域图像的像素与模板图像的对应各点灰度差平方的和来表示的子区域图像与模板图像的相似性与差异性进行模板匹配计算，获取最佳匹配模板和最佳匹配模板在原始MELF元件图像中的最佳匹配位置；其中，T(m,n)表示模板图像T在(m,n)处的灰度值；s(i,j)(m,n)表示与子区域图像s(i,j)中对应的(m,n)处的灰度值；Step 7. Use the template image obtained in step 4 to search for the image to be matched, and the area covered by the template image on the image to be matched is used as a sub-region image, and the template matching calculation formula is used: The template matching calculation is performed on the similarity and difference between the sub-region image and the template image represented by the sum of the squares of the gray level differences between the pixels of the sub-region image and the corresponding points of the template image, and the best matching template and the best matching template are obtained. The best matching position in the original MELF component image; among them, T(m,n) represents the gray value of the template image T at (m,n); s(i,j) (m,n) represents the sub-region The gray value at the corresponding (m, n) in the image s(i, j) ;
步骤七、选取缩小后的旋转角度为0°的模板图像与缩小后的MELF元件图像的距离变换图像进行模板匹配计算，搜索得到最小相关度R(i,j)值，并将最小相关度R(i,j)值对应的位置作为初次粗略匹配的最佳匹配位置，如图8所示；将初次粗略匹配的最佳匹配位置作为矩形的左上顶点，在缩小后的MELF元件图像的距离变换图像中截取一个与缩小后的旋转角度为0°的模板图像相同大小的矩形，再分别向四周扩展相应的偏置量，如图9所示，作为下一步检测图像的感兴趣区域ROI；在获得的感兴趣区域ROI中，用所有的缩小后的旋转角度在[-30°,30°]之间的模板图像进行模板匹配计算，搜索得到第二次粗略匹配相关度值在0.9-1.1倍的最小相关度R(i,j)值内对应的所有匹配模板图像序列号及对应的匹配位置，如图10所示；将获得的所有匹配模板图像形成的候选集中最小相关度R(i,j)值对应的模板图像作为最佳匹配模板图像，如图11所示；使用坐标变换得到最佳匹配模板图像的匹配位置在原始MELF元件图像的距离变换图像中对应的位置，作为矩形的左上顶点，在原始MELF元件图像的距离变换图像中截取一个与最佳匹配序列号对应的原模板图像相同大小的矩形，再分别向四周扩展相应的偏置量，得到如图12所示的感兴趣区域ROI；采用获得的最佳匹配模板图像的最佳匹配序列号，以及最佳匹配序列号前后各三个序列号对应的模板图像，与第二次粗略匹配的最佳匹配位置截取的感兴趣区域ROI和进行模板匹配计算，得到精确匹配的最佳匹配模板图像(如图13所示)的序列号及精确匹配的最佳匹配位置，通过坐标变换得到精确匹配的最佳匹配位置在原始MELF元件图像中对应的位置，如图14所示。Step 7. Select the reduced template image with a rotation angle of 0° and the distance transformed image of the reduced MELF component image for template matching calculation, search to obtain the minimum correlation R (i, j) value, and set the minimum correlation R The position corresponding to the (i, j) value is taken as the best matching position of the initial rough matching, as shown in Figure 8; the best matching position of the first rough matching is taken as the upper left vertex of the rectangle, and the distance transformation of the reduced MELF component image A rectangle with the same size as the reduced template image with a rotation angle of 0° is intercepted from the image, and then the corresponding offset is expanded to the surroundings, as shown in Figure 9, as the next step to detect the ROI of the image; in In the obtained region of interest ROI, use all template images with reduced rotation angles between [-30°, 30°] for template matching calculations, and search to obtain a second rough matching correlation value of 0.9-1.1 times All corresponding matching template image sequence numbers and corresponding matching positions in the minimum correlation R (i, j) value, as shown in Figure 10; the minimum correlation R (i, j) in the candidate set formed by all matching template images obtained j) The template image corresponding to the value is used as the best matching template image, as shown in Figure 11; use coordinate transformation to obtain the matching position of the best matching template image in the corresponding position in the distance transformation image of the original MELF component image, as the upper left of the rectangle At the vertex, a rectangle of the same size as the original template image corresponding to the best matching serial number is intercepted from the distance transformed image of the original MELF component image, and then the corresponding offset is extended to the surroundings respectively, and the interested image shown in Figure 12 is obtained Region ROI; use the best matching serial number of the best matching template image obtained, and the template image corresponding to the three serial numbers before and after the best matching serial number, and the best matching position intercepted from the second rough matching The region ROI and the template matching calculation are performed to obtain the serial number of the best matching template image (as shown in Figure 13) and the best matching position of the exact matching, and the best matching position of the exact matching is obtained by coordinate transformation in the original MELF The corresponding position in the component image is shown in Figure 14.
步骤八、对步骤一获取的原始MELF元件图像采用Canny边缘检测得到带干扰点的MELF边缘图像，利用步骤七得到的最佳匹配模板和最佳匹配模板在原始MELF元件图像中的最佳匹配位置的匹配结果，根据图像边缘点的相关性从带干扰点的MELF边缘图像中提取出关键边缘点，以确定原始MELF图像中MELF元件的整体轮廓，如图15所示；Step 8. Use Canny edge detection on the original MELF component image obtained in step 1 to obtain a MELF edge image with interference points, and use the best matching template obtained in step 7 and the best matching position of the best matching template in the original MELF component image According to the matching result of the image edge point, key edge points are extracted from the MELF edge image with interference points to determine the overall outline of the MELF element in the original MELF image, as shown in Figure 15;
步骤九、利用步骤八获得的所有关键边缘点形成最小外接矩形，用最小外接矩形表示MELF元件在原始MELF元件图像中的位置，最小外接矩形的中心坐标表示MELF元件的中心坐标，最小外接矩形的旋转角度表示MELF元件的旋转角度；在最小外接矩形的中心坐标和旋转角度不变的情况下，将最小外接矩形的长度和宽度放大1.1倍之后形成向外增加偏置量后的矩形；再在最小外接矩形的中心坐标和旋转角度的情况下，将最小外接矩形的长度和宽度缩小0.9倍之后形成向内增加偏置量后的矩形；Step 9: Use all key edge points obtained in step 8 to form a minimum circumscribed rectangle, use the minimum circumscribed rectangle to represent the position of the MELF component in the original MELF component image, the center coordinates of the minimum circumscribed rectangle represent the center coordinates of the MELF component, and the minimum circumscribed rectangle The rotation angle represents the rotation angle of the MELF element; in the case that the center coordinates and rotation angle of the minimum circumscribed rectangle remain unchanged, the length and width of the minimum circumscribed rectangle are enlarged by 1.1 times to form a rectangle with an outwardly increased offset; and then In the case of the center coordinates and rotation angle of the minimum circumscribed rectangle, the length and width of the minimum circumscribed rectangle are reduced by 0.9 times to form a rectangle with an inwardly increased offset;
步骤十、根据步骤九获得的向外增加偏置量后的矩形或向内增加偏置量后的矩形内非零像素的个数，判断检测得到的MELF元件位置是正确的，继续执行步骤十一；Step 10. According to the number of non-zero pixels in the rectangle after increasing the offset amount outward or the rectangle after increasing the offset amount inward obtained in step 9, judge that the detected position of the MELF element is correct, and continue to step 10 one;
步骤十一、判断最小外接矩形的尺寸判断元件的长度和宽度是在容差范围内的，结束MELF元件的定位与检测过程并输出MELF元件的检测结果、中心坐标和旋转角度。Step 11: Judging the size of the minimum circumscribed rectangle, judging that the length and width of the element are within the tolerance range, ending the positioning and detection process of the MELF element and outputting the detection result, center coordinates, and rotation angle of the MELF element.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510358082.1ACN105046271B (en) | 2015-06-25 | 2015-06-25 | The positioning of MELF elements and detection method based on template matches |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510358082.1ACN105046271B (en) | 2015-06-25 | 2015-06-25 | The positioning of MELF elements and detection method based on template matches |
| Publication Number | Publication Date |
|---|---|
| CN105046271Atrue CN105046271A (en) | 2015-11-11 |
| CN105046271B CN105046271B (en) | 2018-03-30 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510358082.1AActiveCN105046271B (en) | 2015-06-25 | 2015-06-25 | The positioning of MELF elements and detection method based on template matches |
| Country | Link |
|---|---|
| CN (1) | CN105046271B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106373123A (en)* | 2016-09-21 | 2017-02-01 | 哈尔滨工业大学 | Surface defect detection method of industrial components based on k_tSL central clustering algorithm |
| CN106485710A (en)* | 2016-10-18 | 2017-03-08 | 广州视源电子科技股份有限公司 | Method and device for detecting wrong component |
| CN106485284A (en)* | 2016-10-19 | 2017-03-08 | 哈尔滨工业大学 | A kind of element localization method based on template matching |
| CN106503720A (en)* | 2016-10-19 | 2017-03-15 | 深圳市路远自动化设备有限公司 | A kind of element image-recognizing method for removing suction nozzle interference |
| CN107436125A (en)* | 2017-08-03 | 2017-12-05 | 环旭电子股份有限公司 | Positioning detection method |
| CN107464240A (en)* | 2017-08-11 | 2017-12-12 | 哈尔滨工业大学 | A kind of location algorithm with rectangular pins chip based on template matches |
| CN107798326A (en)* | 2017-10-20 | 2018-03-13 | 华南理工大学 | A Contour Vision Detection Algorithm |
| CN107909083A (en)* | 2017-11-10 | 2018-04-13 | 中国科学院地理科学与资源研究所 | A kind of hough transform extracting method based on outline optimization |
| CN108257166A (en)* | 2018-01-11 | 2018-07-06 | 上海华虹宏力半导体制造有限公司 | Method for Automatic Matching of Simulation Image of Layout and SEM Image of Silicon Wafer |
| CN108520514A (en)* | 2018-04-04 | 2018-09-11 | 国网上海市电力公司 | Consistency detection method of printed circuit board electronic components based on computer vision |
| CN108890636A (en)* | 2018-07-06 | 2018-11-27 | 陕西大中科技发展有限公司 | A kind of crawl localization method of industrial robot |
| CN109118529A (en)* | 2018-08-13 | 2019-01-01 | 四川长虹电器股份有限公司 | A kind of screw hole Image Quick Orientation method of view-based access control model |
| CN109345460A (en)* | 2018-09-28 | 2019-02-15 | 百度在线网络技术（北京）有限公司 | Method and apparatus for correcting image |
| CN109996077A (en)* | 2019-01-08 | 2019-07-09 | 武汉精立电子技术有限公司 | A kind of logical image decompressing method suitable for display panel detection |
| CN110132823A (en)* | 2019-05-15 | 2019-08-16 | 林伟阳 | A Cell Counting Method Based on Template Matching and Finding Outlines |
| CN110567680A (en)* | 2018-06-05 | 2019-12-13 | 成都精工华耀科技有限公司 | Track fastener looseness detection method based on angle comparison |
| CN110689541A (en)* | 2019-12-09 | 2020-01-14 | 领伟创新智能系统(浙江)有限公司 | Ball screw parameter visual measurement method based on dual-tree complex wavelet |
| CN110689540A (en)* | 2019-11-29 | 2020-01-14 | 领伟创新智能系统(浙江)有限公司 | Ball screw parameter visual measurement method based on image curvature rapid calculation |
| CN111444964A (en)* | 2020-03-27 | 2020-07-24 | 江南大学 | Multi-target rapid image matching method based on self-adaptive ROI (region of interest) division |
| CN112101411A (en)* | 2020-08-05 | 2020-12-18 | 复旦大学 | Welding spot identification and positioning method based on template matching and distribution characteristics |
| CN112419224A (en)* | 2020-07-17 | 2021-02-26 | 宁波智能装备研究院有限公司 | Spherical pin chip positioning method and system |
| CN112509039A (en)* | 2020-12-16 | 2021-03-16 | 广州大学 | Bottled liquid level detection method and system and storage medium |
| CN112712499A (en)* | 2020-12-28 | 2021-04-27 | 合肥联宝信息技术有限公司 | Object detection method and device and computer readable storage medium |
| CN113160148A (en)* | 2021-03-30 | 2021-07-23 | 广东拓斯达科技股份有限公司 | Mold processing method, mold processing device, electronic apparatus, mold processing system, and storage medium |
| CN114003978A (en)* | 2021-11-04 | 2022-02-01 | 三一建筑机器人(西安)研究院有限公司 | Method, device and electronic device for generating workpiece template |
| CN114676229A (en)* | 2022-04-20 | 2022-06-28 | 国网安徽省电力有限公司滁州供电公司 | Technical improvement major repair project file management system and management method |
| WO2022148396A1 (en)* | 2021-01-08 | 2022-07-14 | 长鑫存储技术有限公司 | Collection method for chip, and positioning method for chip |
| CN114898115A (en)* | 2022-05-14 | 2022-08-12 | 上海贝特威自动化科技有限公司 | High-precision template matching method based on local maximum |
| CN115239942A (en)* | 2022-07-26 | 2022-10-25 | 广东工业大学 | Method and equipment for positioning flexible thin piece |
| CN117132797A (en)* | 2023-09-12 | 2023-11-28 | 苏州佳智彩光电科技有限公司 | An LCD display POLMARK positioning detection method and system |
| TWI836744B (en)* | 2022-11-22 | 2024-03-21 | 金寶電子工業股份有限公司 | Image tracking method and image tracking system |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101777118A (en)* | 2010-03-16 | 2010-07-14 | 刘国传 | Method for automatically identifying spots of biochip image on basis of parametric deformable template |
| JP5313080B2 (en)* | 2009-08-18 | 2013-10-09 | クラリオン株式会社 | Linear component reduction device and pedestrian detection display system |
| CN104700085A (en)* | 2015-03-10 | 2015-06-10 | 华中科技大学 | Template matching-based chip positioning method |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5313080B2 (en)* | 2009-08-18 | 2013-10-09 | クラリオン株式会社 | Linear component reduction device and pedestrian detection display system |
| CN101777118A (en)* | 2010-03-16 | 2010-07-14 | 刘国传 | Method for automatically identifying spots of biochip image on basis of parametric deformable template |
| CN104700085A (en)* | 2015-03-10 | 2015-06-10 | 华中科技大学 | Template matching-based chip positioning method |
| Title |
|---|
| ESAKI H等: "Automatic generation of IC component configuration data", 《ELECTRICAL ENGINEERING IN JAPAN》* |
| 周丽莎: "基于模板匹配的视觉定位技术研究与应用", 《中国优秀硕士学位论文全文数据库 信息科技辑》* |
| 李铁军等: "QUAD类芯片的视觉检测与定位技术研究", 《仪器仪表学报》* |
| 麦倩: "基于配准的新型表面贴装元器件的定位算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》* |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106373123B (en)* | 2016-09-21 | 2019-01-08 | 哈尔滨工业大学 | Surface defect detection method of industrial components based on k_tSL center clustering algorithm |
| CN106373123A (en)* | 2016-09-21 | 2017-02-01 | 哈尔滨工业大学 | Surface defect detection method of industrial components based on k_tSL central clustering algorithm |
| CN106485710A (en)* | 2016-10-18 | 2017-03-08 | 广州视源电子科技股份有限公司 | Method and device for detecting wrong component |
| CN106485284A (en)* | 2016-10-19 | 2017-03-08 | 哈尔滨工业大学 | A kind of element localization method based on template matching |
| CN106503720A (en)* | 2016-10-19 | 2017-03-15 | 深圳市路远自动化设备有限公司 | A kind of element image-recognizing method for removing suction nozzle interference |
| CN106503720B (en)* | 2016-10-19 | 2019-09-13 | 深圳市路远自动化设备有限公司 | A kind of element image-recognizing method of removal suction nozzle interference |
| CN107436125A (en)* | 2017-08-03 | 2017-12-05 | 环旭电子股份有限公司 | Positioning detection method |
| CN107436125B (en)* | 2017-08-03 | 2020-03-10 | 环旭电子股份有限公司 | Positioning detection method |
| CN107464240B (en)* | 2017-08-11 | 2020-07-24 | 哈尔滨工业大学 | A Location Algorithm for Chips with Rectangular Pins Based on Template Matching |
| CN107464240A (en)* | 2017-08-11 | 2017-12-12 | 哈尔滨工业大学 | A kind of location algorithm with rectangular pins chip based on template matches |
| CN107798326A (en)* | 2017-10-20 | 2018-03-13 | 华南理工大学 | A Contour Vision Detection Algorithm |
| CN107909083A (en)* | 2017-11-10 | 2018-04-13 | 中国科学院地理科学与资源研究所 | A kind of hough transform extracting method based on outline optimization |
| CN107909083B (en)* | 2017-11-10 | 2019-03-15 | 中国科学院地理科学与资源研究所 | A kind of hough transform extracting method based on outline optimization |
| CN108257166B (en)* | 2018-01-11 | 2022-03-04 | 上海华虹宏力半导体制造有限公司 | Method for automatically matching simulation image of layout with silicon wafer SEM image |
| CN108257166A (en)* | 2018-01-11 | 2018-07-06 | 上海华虹宏力半导体制造有限公司 | Method for Automatic Matching of Simulation Image of Layout and SEM Image of Silicon Wafer |
| CN108520514B (en)* | 2018-04-04 | 2020-05-12 | 国网上海市电力公司 | Consistency detection method of printed circuit board electronic components based on computer vision |
| CN108520514A (en)* | 2018-04-04 | 2018-09-11 | 国网上海市电力公司 | Consistency detection method of printed circuit board electronic components based on computer vision |
| CN110567680A (en)* | 2018-06-05 | 2019-12-13 | 成都精工华耀科技有限公司 | Track fastener looseness detection method based on angle comparison |
| CN110567680B (en)* | 2018-06-05 | 2022-03-29 | 成都精工华耀科技有限公司 | Track fastener looseness detection method based on angle comparison |
| CN108890636A (en)* | 2018-07-06 | 2018-11-27 | 陕西大中科技发展有限公司 | A kind of crawl localization method of industrial robot |
| CN109118529B (en)* | 2018-08-13 | 2022-06-03 | 四川长虹电器股份有限公司 | Screw hole image rapid positioning method based on vision |
| CN109118529A (en)* | 2018-08-13 | 2019-01-01 | 四川长虹电器股份有限公司 | A kind of screw hole Image Quick Orientation method of view-based access control model |
| CN109345460A (en)* | 2018-09-28 | 2019-02-15 | 百度在线网络技术（北京）有限公司 | Method and apparatus for correcting image |
| CN109996077A (en)* | 2019-01-08 | 2019-07-09 | 武汉精立电子技术有限公司 | A kind of logical image decompressing method suitable for display panel detection |
| CN110132823A (en)* | 2019-05-15 | 2019-08-16 | 林伟阳 | A Cell Counting Method Based on Template Matching and Finding Outlines |
| CN110689540A (en)* | 2019-11-29 | 2020-01-14 | 领伟创新智能系统(浙江)有限公司 | Ball screw parameter visual measurement method based on image curvature rapid calculation |
| CN110689540B (en)* | 2019-11-29 | 2020-04-07 | 领伟创新智能系统(浙江)有限公司 | Ball screw parameter visual measurement method based on image curvature rapid calculation |
| CN110689541A (en)* | 2019-12-09 | 2020-01-14 | 领伟创新智能系统(浙江)有限公司 | Ball screw parameter visual measurement method based on dual-tree complex wavelet |
| CN110689541B (en)* | 2019-12-09 | 2020-05-08 | 领伟创新智能系统(浙江)有限公司 | Ball screw parameter visual measurement method based on dual-tree complex wavelet |
| CN111444964B (en)* | 2020-03-27 | 2023-08-08 | 江南大学 | Multi-target rapid image matching method based on adaptive ROI (region of interest) division |
| CN111444964A (en)* | 2020-03-27 | 2020-07-24 | 江南大学 | Multi-target rapid image matching method based on self-adaptive ROI (region of interest) division |
| CN112419224B (en)* | 2020-07-17 | 2021-08-27 | 宁波智能装备研究院有限公司 | Spherical pin chip positioning method and system |
| CN112419224A (en)* | 2020-07-17 | 2021-02-26 | 宁波智能装备研究院有限公司 | Spherical pin chip positioning method and system |
| CN112101411A (en)* | 2020-08-05 | 2020-12-18 | 复旦大学 | Welding spot identification and positioning method based on template matching and distribution characteristics |
| CN112101411B (en)* | 2020-08-05 | 2022-08-19 | 复旦大学 | Welding spot identification and positioning method based on template matching and distribution characteristics |
| CN112509039A (en)* | 2020-12-16 | 2021-03-16 | 广州大学 | Bottled liquid level detection method and system and storage medium |
| CN112712499A (en)* | 2020-12-28 | 2021-04-27 | 合肥联宝信息技术有限公司 | Object detection method and device and computer readable storage medium |
| CN112712499B (en)* | 2020-12-28 | 2022-02-01 | 合肥联宝信息技术有限公司 | Object detection method and device and computer readable storage medium |
| WO2022148396A1 (en)* | 2021-01-08 | 2022-07-14 | 长鑫存储技术有限公司 | Collection method for chip, and positioning method for chip |
| US11861451B2 (en) | 2021-01-08 | 2024-01-02 | Changxin Memory Technologies, Inc. | Method for chip collection and method for chip positioning |
| CN113160148A (en)* | 2021-03-30 | 2021-07-23 | 广东拓斯达科技股份有限公司 | Mold processing method, mold processing device, electronic apparatus, mold processing system, and storage medium |
| WO2022205606A1 (en)* | 2021-03-30 | 2022-10-06 | 广东拓斯达科技股份有限公司 | Mould processing method and apparatus, electronic device, system, and storage medium |
| CN114003978A (en)* | 2021-11-04 | 2022-02-01 | 三一建筑机器人(西安)研究院有限公司 | Method, device and electronic device for generating workpiece template |
| CN114676229A (en)* | 2022-04-20 | 2022-06-28 | 国网安徽省电力有限公司滁州供电公司 | Technical improvement major repair project file management system and management method |
| CN114898115A (en)* | 2022-05-14 | 2022-08-12 | 上海贝特威自动化科技有限公司 | High-precision template matching method based on local maximum |
| CN115239942A (en)* | 2022-07-26 | 2022-10-25 | 广东工业大学 | Method and equipment for positioning flexible thin piece |
| TWI836744B (en)* | 2022-11-22 | 2024-03-21 | 金寶電子工業股份有限公司 | Image tracking method and image tracking system |
| CN117132797A (en)* | 2023-09-12 | 2023-11-28 | 苏州佳智彩光电科技有限公司 | An LCD display POLMARK positioning detection method and system |
| Publication number | Publication date |
|---|---|
| CN105046271B (en) | 2018-03-30 |
| Publication | Publication Date | Title |
|---|---|---|
| CN105046271A (en) | MELF (Metal Electrode Leadless Face) component positioning and detecting method based on match template | |
| CN104200461B (en) | The remote sensing image registration method of block and sift features is selected based on mutual information image | |
| CN104318548B (en) | Rapid image registration implementation method based on space sparsity and SIFT feature extraction | |
| CN105335973B (en) | Apply to the visual processing method of strip machining production line | |
| CN106446894B (en) | A method of based on outline identification ball-type target object location | |
| CN104574421B (en) | A method and device for high-precision multi-spectral image registration in large format and small overlapping areas | |
| CN103136525B (en) | High-precision positioning method for special-shaped extended target by utilizing generalized Hough transformation | |
| CN105184770B (en) | A kind of soldered ball positioning and its parameter identification method for ball grid array pin chip | |
| CN104809446B (en) | Palmmprint area-of-interest rapid extracting method based on correction volar direction | |
| CN103778626B (en) | A kind of fast image registration method of view-based access control model marking area | |
| CN111126174A (en) | Visual detection method for robot to grab parts | |
| CN112396643A (en) | Multi-mode high-resolution image registration method with scale-invariant features and geometric features fused | |
| CN103020945A (en) | Remote sensing image registration method of multi-source sensor | |
| CN106296587B (en) | Tire mold image stitching method | |
| CN101556647A (en) | mobile robot visual orientation method based on improved SIFT algorithm | |
| CN101159017A (en) | A Vision Method for Automatic Weld Seam Recognition Based on Local Image Texture Feature Matching | |
| CN104252711A (en) | Matching method based on image gradient direction code | |
| CN106485749A (en) | A kind of rectangular pins element rough localization method based on angle point | |
| CN104077775A (en) | Shape matching method and device combining skeleton feature points and shape context | |
| CN104966047A (en) | Method and device for identifying vehicle license | |
| CN111860501A (en) | Image recognition method of high-speed rail height adjustment rod falling out fault based on shape matching | |
| CN103761768A (en) | Stereo matching method of three-dimensional reconstruction | |
| CN103247032B (en) | Weak extended target positioning method based on attitude compensation | |
| CN110222661A (en) | It is a kind of for motion estimate and the feature extracting method of tracking | |
| CN111667429A (en) | Target positioning and correcting method for inspection robot |
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right | Effective date of registration:20190912 Address after:150001 No. 434, postal street, Nangang District, Heilongjiang, Harbin Co-patentee after:Gao Huijun Patentee after:Harbin Institute of Technology Asset Investment Management Co.,Ltd. Address before:150001 Harbin, Nangang, West District, large straight street, No. 92 Patentee before:Harbin Institute of Technology | |
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right | Effective date of registration:20191021 Address after:315200 No.189, Guangming Road, Zhuangshi street, Zhenhai District, Ningbo City, Zhejiang Province Patentee after:Ningbo Intelligent Equipment Research Institute Co.,Ltd. Address before:150001 No. 434, postal street, Nangang District, Heilongjiang, Harbin Co-patentee before:Gao Huijun Patentee before:Harbin Institute of Technology Asset Investment Management Co.,Ltd. | |
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right | Effective date of registration:20250303 Address after:No. 289 Dongsheng Road, Jiaochuan Street, Zhenhai District, Ningbo City, Zhejiang Province 315200 Patentee after:Ningbo Yitang Intelligent Technology Co.,Ltd. Country or region after:China Address before:315200 No.189 Guangming Road, Zhuangshi street, Zhenhai District, Ningbo City, Zhejiang Province Patentee before:Ningbo Intelligent Equipment Research Institute Co.,Ltd. Country or region before:China |