技术领域technical field
本发明涉及一种测距方法，尤其涉及一种基于双目合成孔径聚焦图像的目标测距方法、装置。The invention relates to a ranging method, in particular to a target ranging method and device based on a binocular synthetic aperture focusing image.
背景技术Background technique
测量目标距离的方法有多种，如基于激光脉冲时间差的点测距法、基于幅度调制波相位差的深度获取、几何光学聚焦法、Moire拓扑技术、全息干涉测量法、Fresnel衍射技术和结构光法等。超声也常用来获取深度信息。计算机立体视觉技术近年来在许多领域得到广泛应用。基本方法是从两个或多个视点去观察同一场景，获得在不同视角下的一组图像，获得不同图像中对应像素间的视差，然后通过三角计算测量出场景中目标的深度信息,它需要确定双目或多目图像中的对应点，这是一个很困难的问题。当空间三维场景被投影为二维图像时一些有用信息由于投影而丢失了，同一景物在不同视点下的图像中会有很大的不同，受遮挡或阴影的影响，景物的若干点有可能不出现在所有图像中，而且场景中的诸多变化因素，如光照条件、噪声干扰、景物几何形状的畸变、表面物理特性以及摄像机特性等，都被综合到单一的图像灰度值中，要仅由这一灰度值确定以上诸多因素是十分困难的，至今这个问题还没有得到很好的解决。增大基线长度可以改善深度测量精度，但同时会增大图像间的差异，增加匹配的困难程度。多机位小孔成像可合成为大孔径图像，与单镜头成像一致。日本有文献[Kusumoto,N.；Hiura,S.；Sato,K.UncalibratedSyntheticApertureforDefocusControlIEEEConferenceonComputerVisionandPatternRecognition,2009.CVPR2009.P:2252-2259]研究了利用合成孔径方法对针孔成像照片进行艺术加工，使之产生非主体散焦效果，以突出视觉重点。单目聚/散焦测距法发展较为成熟，根据合成孔径原理使用多个摄像机可获得单目大孔径效应，因此可借鉴单目成像丰富的聚/散焦测距算法，使得合成孔径成像测距拥有较高的发展基础。由于符合针孔成像特点的数码设备造价很低，成像结果便于数字化处理，合成孔径成像的发展前景广阔。There are many methods to measure the target distance, such as point ranging method based on laser pulse time difference, depth acquisition based on amplitude modulated wave phase difference, geometric optics focusing method, Moire topology technology, holographic interferometry, Fresnel diffraction technology and structured light law etc. Ultrasound is also commonly used to obtain depth information. Computer stereo vision technology has been widely used in many fields in recent years. The basic method is to observe the same scene from two or more viewpoints, obtain a set of images under different viewpoints, obtain the parallax between corresponding pixels in different images, and then measure the depth information of the target in the scene through triangulation calculation. Corresponding points in binocular or multi-view images need to be determined, which is a difficult problem. When the spatial 3D scene is projected into a 2D image, some useful information is lost due to the projection. The images of the same scene at different viewpoints will be very different. Affected by occlusion or shadows, some points of the scene may not be correct. Appears in all images, and many changing factors in the scene, such as lighting conditions, noise interference, distortion of scene geometry, surface physical characteristics and camera characteristics, etc., are integrated into a single gray value of the image. It is very difficult to determine the gray value of the above factors, and this problem has not been well resolved so far. Increasing the baseline length can improve the accuracy of depth measurement, but at the same time it will increase the difference between images and increase the difficulty of matching. Multi-camera small-aperture imaging can be synthesized into a large-aperture image, which is consistent with single-lens imaging. Japanese literature [Kusumoto, N.; Hiura, S.; Sato, K. Uncalibrated Synthetic Aperture for Defocus Control IEEE Conference on Computer Vision and Pattern Recognition, 2009. CVPR2009.P: 2252-2259] studied the artistic processing of pinhole imaging photos by using synthetic aperture method to make them non-subject Defocus effect to emphasize visual focus. The monocular focusing/defocusing distance measurement method is relatively mature. According to the principle of synthetic aperture, the monocular large aperture effect can be obtained by using multiple cameras. It has a high development base. Because the cost of digital equipment that conforms to the characteristics of pinhole imaging is very low, and the imaging results are easy to digitally process, the development prospect of synthetic aperture imaging is broad.
一篇中国发明专利(CN103033166B)公开了一种基于合成孔径聚焦图像的目标测距方法，该方法包括以下步骤：步骤1、利用小孔成像模型摄像机获取与目标视线垂直的等间隔线阵机位图像序列，线阵与目标视线交点处机位的图像作为基准图像；步骤2、将可测距离范围分成多个距离段，对于每一个距离段，先分别计算出所述图像序列中各幅图像与基准图像之间的像差，然后将图像序列中各幅图像进行像差校正后进行叠加，得到该距离段所对应的像差校正叠加图像；每一个距离段对应一幅像差校正叠加图像；A Chinese invention patent (CN103033166B) discloses a target ranging method based on a synthetic aperture focusing image, which includes the following steps: Step 1, using a pinhole imaging model camera to obtain an equidistant line array position perpendicular to the target line of sight Image sequence, the image of the machine position at the intersection of the line array and the target line of sight is used as the reference image; step 2, the measurable distance range is divided into multiple distance segments, and for each distance segment, firstly calculate each image in the image sequence Then the images in the image sequence are aberration corrected and superimposed to obtain the aberration corrected superimposed image corresponding to the distance segment; each distance segment corresponds to an aberration corrected superimposed image ;
步骤3、计算基准图像中每个像素的邻域与每一幅像差校正叠加图像中相应区域的相似度，并选取相似度随像差校正叠加图像变化的范围大于一预设阈值的像素作为可测距像素；步骤4、对于基准图像中的每一个可测距像素，选出相应区域与该可测距像素的邻域的相似度最大的像差校正叠加图像，该像差校正叠加图像所对应的距离段即为该可测距像素对应目标点所处的距离段。该方法利用利用合成孔径聚焦成像的原理进行目标测距，具有实现成本低、抗干扰能力强、算法简单等优点。然而，获得精确的合成孔径聚焦像是困难的，尤其是大孔径，移动拍摄图像序列的聚焦更难；散焦像斑块的干扰也不容忽视；此外，合成孔径叠加像具有超分辨率重建潜力，而现有单目测距对此潜力难以利用。Step 3. Calculate the similarity between the neighborhood of each pixel in the reference image and the corresponding area in each aberration-corrected superimposed image, and select pixels whose similarity varies with the aberration-corrected superimposed image to a range greater than a preset threshold as Range-measurable pixels; step 4, for each range-measurable pixel in the reference image, select the aberration-corrected superimposed image with the largest similarity between the corresponding area and the neighborhood of the range-measurable pixel, and the aberration-corrected superimposed image The corresponding distance segment is the distance segment where the distance-measurable pixel corresponds to the target point. This method uses the principle of synthetic aperture focusing imaging to measure the distance of the target, and has the advantages of low cost, strong anti-interference ability, and simple algorithm. However, it is difficult to obtain accurate synthetic aperture focusing images, especially for large apertures, and it is even more difficult to focus on moving image sequences; the interference of defocused image patches cannot be ignored; in addition, synthetic aperture stacked images have the potential for super-resolution reconstruction , and the existing monocular ranging is difficult to utilize this potential.
发明内容Contents of the invention
本发明所要解决的技术问题在于克服现有技术不足，提供一种基于双目合成孔径聚焦图像的目标测距方法，可有效抑制共模误差和散焦像干扰，改善合成孔径聚焦图像测距的性能。The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide a target ranging method based on binocular synthetic aperture focusing images, which can effectively suppress common mode errors and defocus image interference, and improve the performance of synthetic aperture focusing image ranging. performance.
本发明具体采用以下技术方案解决上述技术问题：The present invention specifically adopts the following technical solutions to solve the above technical problems:
一种基于双目合成孔径聚焦图像的目标测距方法，包括以下步骤：A target ranging method based on binocular synthetic aperture focusing images, comprising the following steps:
步骤1、从两个结构相同的面阵处，以所述两个面阵的对称中心作为基准机位，对目标所在场景分别进行光学合成孔径成像，分别得到一组对应不同聚焦距离的合成孔径图像；Step 1. From two area arrays with the same structure, take the symmetry center of the two area arrays as the reference camera position, perform optical synthetic aperture imaging on the scene where the target is located, and obtain a set of synthetic apertures corresponding to different focusing distances image;
步骤2、对每一个聚焦距离，将其所对应的从两个面阵处获得的两幅合成孔径图像作为一对，然后对每一对合成孔径图像进行以下处理：以其中一幅图像中的目标图像作为参考图像，将另一幅图像中与参考图像相对应的区域作为匹配位置图像，并将另一幅图像中与匹配位置图像间的像差矢量的模在(0,d]范围内变化的一系列区域作为非匹配位置图像，d为预设的最大像差；计算参考图像与匹配位置图像之间的相容信息量R0以及参考图像与各非匹配位置图像之间的相容信息量，并将所计算出的参考图像与各非匹配位置图像之间的相容信息量中大于R0的值统一修改为R0，然后计算修改后的参考图像与各非匹配位置图像之间的相容信息量的均值；最后以R0与所述均值的差值作为该聚焦距离的检测量；所述相容信息量用于度量两幅图像的信息中相同部分的量；Step 2. For each focusing distance, take the corresponding two synthetic aperture images obtained from the two area arrays as a pair, and then perform the following processing on each pair of synthetic aperture images: The target image is used as a reference image, and the area corresponding to the reference image in the other image is used as the matching position image, and the modulus of the aberration vector between the other image and the matching position image is in the range of (0, d] A series of changing areas are used as non-matching position images, d is the preset maximum aberration; calculate the compatible information amount R0 between the reference image and the matching position image and the compatibility between the reference image and each non-matching position image Information amount, and uniformly modify the value greater than R0 in the calculated compatible information amount between the reference image and each non-matching position image to R0 , and then calculate the difference between the modified reference image and each non-matching position image The average value of the compatible information amount between; finally, the difference between R0 and the average value is used as the detection amount of the focus distance; the compatible information amount is used to measure the amount of the same part of the information of the two images;
步骤3、找出检测量最大的聚焦距离，并判断该聚焦距离所对应的R0是否大于预设的相容信息量门限值，如是，则以该聚焦距离作为目标距离。Step 3. Find out the focus distance with the largest detection amount, and judge whether the R0 corresponding to the focus distance is greater than the preset compatible information amount threshold, and if so, use the focus distance as the target distance.
根据相同的发明思路还可以得到以下技术方案：According to the same inventive idea, the following technical solutions can also be obtained:
一种基于双目合成孔径聚焦图像的目标测距装置，包括：A target ranging device based on a binocular synthetic aperture focusing image, comprising:
双目合成孔径成像单元，用于从两个结构相同的面阵处，以所述两个面阵的对称中心作为基准机位，对目标所在场景分别进行光学合成孔径成像，分别得到一组对应不同聚焦距离的合成孔径图像；The binocular synthetic aperture imaging unit is used to perform optical synthetic aperture imaging on the scene where the target is located from two planar arrays with the same structure, using the symmetry center of the two planar arrays as a reference position, and obtain a set of corresponding Synthetic aperture images at different focus distances;
距离检测单元，用于根据双目合成孔径成像单元所获得的两组对应不同聚焦距离的合成孔径图像检测出目标距离，其检测方法具体如下：对每一个聚焦距离，将其所对应的从两个面阵处获得的两幅合成孔径图像作为一对，然后对每一对合成孔径图像进行以下处理：以其中一幅图像中的目标图像作为参考图像，将另一幅图像中与参考图像相对应的区域作为匹配位置图像，并将另一幅图像中与匹配位置图像间的像差矢量的模在(0,d]范围内变化的一系列区域作为非匹配位置图像，d为预设的最大像差；计算参考图像与匹配位置图像之间的相容信息量R0以及参考图像与各非匹配位置图像之间的相容信息量，并将所计算出的参考图像与各非匹配位置图像之间的相容信息量中大于R0的值统一修改为R0，然后计算修改后的参考图像与各非匹配位置图像之间的相容信息量的均值；最后以R0与所述均值的差值作为该聚焦距离的检测量；所述相容信息量用于度量两幅图像的信息中相同部分的量；找出检测量最大的聚焦距离，并判断该聚焦距离所对应的R0是否大于预设的相容信息量门限值，如是，则以该聚焦距离作为目标距离。The distance detection unit is used to detect the target distance according to the two sets of synthetic aperture images corresponding to different focusing distances obtained by the binocular synthetic aperture imaging unit. The two synthetic aperture images obtained at the area arrays are regarded as a pair, and then each pair of synthetic aperture images is processed as follows: the target image in one image is used as a reference image, and the other image is compared with the reference image The corresponding area is used as the matching position image, and a series of areas in which the modulus of the disparity vector between the matching position image and the other image changes in the range of (0, d] is taken as the non-matching position image, and d is the preset Maximum aberration; calculate the compatible information amount R0 between the reference image and the matching position image and the compatible information amount between the reference image and each non-matching position image, and combine the calculated reference image with each non-matching position The values greater than R0 in the compatible information between images are uniformly modified to R0 , and then the average value of the compatible information between the modified reference image and each non-matching position image is calculated; finally, R0 and the described The difference of the average value is used as the detection amount of the focus distance; the amount of compatible information is used to measure the amount of the same part of the information of the two images; find the focus distance with the maximum detection amount, and judge the R value corresponding to the focus distance0 is greater than the preset compatible information threshold, if yes, the focus distance is used as the target distance.
优选地，所述面阵为圆形面阵；所述光学合成孔径成像通过密布在圆形面阵中的小孔成像模型摄像机阵列同时拍摄实现，或者通过单个小孔成像模型摄像机在圆形面阵中一系列密集排布的机位分时拍摄实现。Preferably, the area array is a circular area array; the optical synthetic aperture imaging is realized by shooting simultaneously with a small hole imaging model camera array densely distributed in the circular area array, or by a single small hole imaging model camera on a circular surface Time-sharing shooting of a series of densely arranged camera positions in the array is realized.
优选地，对于灰度图像，所述相容信息量的度量形式为两幅图像的信息熵与两幅图像间相似度的乘积，或者为两幅图像的平均信息熵与两幅图像间相似度的乘积，或者为两幅图像的锐度均值或梯度均值或对比度均值与两幅图像间相似度的乘积，或者为两幅图像的互熵；对于彩色图像，所述相容信息量的度量形式为采用灰度图像的相容信息量度量形式所得到的两幅图像在RGB颜色空间中各通道子图像间相容信息量的和或均值。Preferably, for a grayscale image, the metric form of the compatible information amount is the product of the information entropy of the two images and the similarity between the two images, or the average information entropy of the two images and the similarity between the two images The product of , or the product of the sharpness mean or gradient mean or contrast mean of the two images and the similarity between the two images, or the cross-entropy of the two images; for color images, the measure of the amount of compatible information It is the sum or mean value of the consistent information content between the sub-images of each channel in the RGB color space of the two images obtained by using the compatible information content measurement form of the grayscale image.
优选地，所述像差矢量的模在(0,d]范围内逐像素变化。Preferably, the modulus of the aberration vector varies pixel by pixel within the range (0,d].
相比现有技术，本发明具有以下有益效果：Compared with the prior art, the present invention has the following beneficial effects:
1)用信号处理方法能实现成像面的灵活移动，镜头孔径(光圈)灵活张缩，用信号处理方法能实现聚/散测量距离信息。1) The flexible movement of the imaging surface can be realized by the signal processing method, and the lens aperture (aperture) can be flexibly expanded and contracted, and the convergence/divergence measurement distance information can be realized by the signal processing method.
2)该方法抛弃了图像匹配的过程，而根据所有图像信息获得目标距离，结果受单个摄像机影响极小，也基本不受图像内容影响。2) This method abandons the process of image matching, and obtains the target distance based on all image information. The result is minimally affected by a single camera, and is basically not affected by image content.
3)可利用同一小孔成像模型摄像机分时拍摄得到双目合成孔径图像序列,便于实现，尤其是对于航拍测距的应用更具有重要意义。3) The binocular synthetic aperture image sequence can be obtained by time-sharing shooting with the same pinhole imaging model camera, which is easy to implement, especially for the application of aerial distance measurement.
4)不向地面发射信号的被动测绘具有隐蔽，无电磁波污染，能有效降低测绘成本；4) Passive surveying and mapping that does not transmit signals to the ground is concealed and has no electromagnetic wave pollution, which can effectively reduce the cost of surveying and mapping;
5)左右面阵密布在两个大圆面阵内，合成孔径长度较小，可避免桶型失真校正，以实现实时测量。5) The left and right arrays are densely distributed in two large circular arrays, and the synthetic aperture length is small, which can avoid barrel distortion correction and realize real-time measurement.
6)相比现有基于合成孔径聚焦图像的目标测距方法，本发明可抑制共模误差和散焦像干扰，改善了合成孔径聚焦图像测距的性能。6) Compared with the existing target ranging method based on the synthetic aperture focused image, the present invention can suppress the common mode error and defocus image interference, and improve the performance of the synthetic aperture focused image ranging.
7)因聚散焦效应，不在聚焦距离的目标信息熵被衰减，图像信息熵更容易来自同一距离的目标，因此合成孔径叠加像中目标邻域可以相对大，以提高测距信息来源。7) Due to the defocusing effect, the information entropy of targets not at the focus distance is attenuated, and the image information entropy is more likely to come from targets at the same distance. Therefore, the target neighborhood in the synthetic aperture superimposed image can be relatively large to improve the source of ranging information.
附图说明Description of drawings
图1为光学合成孔径成像原理示意图；Figure 1 is a schematic diagram of the principle of optical synthetic aperture imaging;
图2为本发明具体实施方式中的目标测距方法流程示意图；Fig. 2 is a schematic flow chart of a target ranging method in a specific embodiment of the present invention;
图3为本发明优选使用的双目合成孔径成像单元示意图；Fig. 3 is a schematic diagram of a binocular synthetic aperture imaging unit preferably used in the present invention;
图4为小孔摄像机分时采集到的图像序列；Fig. 4 is the image sequence collected by the pinhole camera in time-sharing;
图5为所得到的部分合成孔径图像，其中(a)为左目聚焦在车辆前沿处，(b)为左目聚焦在相对远景处，(c)为右目聚焦在车辆前沿处，(d)为右目聚焦在相对远景处；Figure 5 is part of the obtained synthetic aperture images, where (a) is the left eye focusing on the front of the vehicle, (b) is the left eye focusing on the relative distance, (c) is the right eye focusing on the front of the vehicle, (d) is the right eye Focus on the relative distance;
图6为不同目标的双目目标图像相容信息量随聚焦距离和视差的变化分布示意图；其中，(a)的测距目标在建筑边沿线上，(b)为(a)所对应的相容信息量随聚焦距离和视差的变化分布示意图，(c)的测距目标在建筑边沿线以上部分，(d)为(c)所对应的相容信息量随聚焦距离和视差的变化分布示意图，(e)为经削峰处理后的相容信息量随聚焦距离和视差的变化分布示意图；Figure 6 is a schematic diagram of the distribution of binocular target image compatible information of different targets with the focus distance and parallax; where, (a) the distance measurement target is on the building edge, (b) is the corresponding phase of (a) Schematic diagram of the distribution of the content of information with the focus distance and parallax, (c) the distance measurement target is above the building edge, (d) is a schematic diagram of the distribution of the content of content with the focus distance and parallax corresponding to (c) , (e) is a schematic diagram of the distribution of the amount of compatible information with the focus distance and parallax after peak clipping processing;
图7为检测量随距离变化曲线，其中，(a)为测距目标在建筑边沿线上时相容信息量随聚焦距离的变化曲线，(b)为测距目标在建筑边沿线以上部分时相容信息量随聚焦距离的变化曲线，(c)为测距目标在建筑边沿线以上部分时双目匹配位置相容信息量与非匹配位置相容信息量均值之差随聚焦距离的变化曲线；Figure 7 is the variation curve of the detection amount with the distance, where (a) is the variation curve of the compatible information amount with the focusing distance when the ranging target is on the building edge line, and (b) is when the ranging target is above the building edge line The change curve of the compatible information amount with the focusing distance, (c) is the change curve of the difference between the average compatible information amount of the binocular matching position and the non-matching position with the focusing distance when the ranging target is above the building edge ;
图8为利用本发明方法获得的目标深度图像。Fig. 8 is a target depth image obtained by using the method of the present invention.
具体实施方式detailed description
下面结合附图对本发明的技术方案进行详细说明：The technical scheme of the present invention is described in detail below in conjunction with accompanying drawing:
为便于公众理解，首先对合成孔径聚焦成像的基本原理进行介绍。大孔径镜头成像时，通过镜头各部位的入射光线经镜头折射后，目标光线重新会聚，在成像面成二维像。若将入射光线按所通过的子孔径分离成像，再用信号处理方法叠加，就得到合成孔径像，成像结果与大孔径镜头成像一致。如图1所示，全孔径成像时，正确距离处点经大孔径光学镜头聚焦在成像面上，为一点；而另一距离处的目标点成像为一圆，它由各子孔径成像组成，因此我们能仿真大孔径镜头成像过程，可以应用单目测距的方法测量合成孔径目标的距离。形成合成孔径像，并没有要求图1大孔径圈中所有子孔径像都参与计算，只需部分子孔径像参与叠加计算，就能形成合成孔径效应。合成孔径测距与一般多目视觉的区别在于合成孔径聚焦形成的合成孔径增益，提高了信噪比，此外它对特定距离聚焦的同时，其他距离处的目标像散焦为均匀平滑的像斑，这些像斑对测距的干扰相对较小，也就是说合成孔径通过散焦滤除了其他距离处目标像干扰。合成孔径聚焦测距法只对特定距离目标聚焦，聚焦目标信号能量集中，受其他距离目标信号影响小，聚焦像能量集中在小区域，易于与其他距离目标区分，算法稳定，适于工程应用。然而当合成孔径长度较大时，获得聚焦清晰的像是困难的，桶型失真等因素使得通用的聚焦校正十分复杂。In order to facilitate the public's understanding, the basic principle of synthetic aperture focusing imaging is firstly introduced. When imaging with a large-aperture lens, the incident light rays that pass through various parts of the lens are refracted by the lens, and the target light rays converge again to form a two-dimensional image on the imaging surface. If the incident light is separated and imaged according to the sub-apertures it passes through, and then superimposed by the signal processing method, a synthetic aperture image is obtained, and the imaging result is consistent with that of a large-aperture lens. As shown in Figure 1, during full-aperture imaging, the point at the correct distance is focused on the imaging surface by the large-aperture optical lens, which is a point; while the target point at another distance is imaged as a circle, which is composed of sub-aperture images. Therefore, we can simulate the imaging process of the large-aperture lens, and can use the method of monocular distance measurement to measure the distance of the synthetic aperture target. To form a synthetic aperture image, it is not required that all the sub-aperture images in the large-aperture circle in Figure 1 participate in the calculation, and only part of the sub-aperture images need to participate in the superposition calculation to form the synthetic aperture effect. The difference between synthetic aperture ranging and general multi-eye vision lies in the synthetic aperture gain formed by synthetic aperture focusing, which improves the signal-to-noise ratio. In addition, it focuses on a specific distance while defocusing the target at other distances into a uniform and smooth image spot , the interference of these image spots on ranging is relatively small, that is to say, the synthetic aperture filters out the interference of target images at other distances through defocusing. The synthetic aperture focusing distance measurement method only focuses on a target at a specific distance, and the energy of the focused target signal is concentrated, which is less affected by the signals of other distant targets. The energy of the focused image is concentrated in a small area, which is easy to distinguish from other distance targets. The algorithm is stable and suitable for engineering applications. However, when the synthetic aperture length is large, it is difficult to obtain a sharply focused image, and factors such as barrel distortion make general focus correction very complicated.
大孔径镜头只对特定距离目标聚焦，而其他距离处目标散焦。目标信息以图像的熵的形式衡量，聚焦像保持了目标图像的信息熵，而散焦效应则使目标信息衰减，因此可以通过检测目标图像信息熵的保留状况判断目标距离。目标图像信息熵的保留状况理论上可用左、右目合成孔径图像间的相容信息量衡量，这正是双目合成孔径聚焦测距的基本方法。两幅图像间的相容信息量是指两幅图像的信息中相同部分的量,其具体的度量可采用各种合理的形式，优选地，对于灰度图像，所述相容信息量的度量形式为两幅图像的信息熵与两幅图像间相似度的乘积，或者为两幅图像的平均信息熵与两幅图像间相似度的乘积，或者两幅图像的互熵；对于彩色图像，所述相容信息量的度量形式为采用灰度图像的相容信息量度量形式所得到的两幅图像在RGB颜色空间中各通道子图像间相容信息量的和或均值。当然，也可采用其他度量形式，例如可利用两幅图像的锐度或梯度或对比度代替上述度量形式中的信息熵，同样可实现相容信息量的度量；也可以采用或自行构造与上述相容信息量具有相同变化趋势的度量形式。Studholme提出了归一化互信息(NormalizedMutualInformation,NMI)：式中(x,y)为像素坐标，H(R)和H(F)别是基准图像和合成图像在坐标(x,y)邻域图像块的信息熵，H(R,F)联合熵，最常用直方图法计算，从直观的角度突出了以频率代替概率来进行密度估计，本具体实施方式即采用直方图法计算互信息熵来度量相容信息量。利用单目合成孔径聚焦图像进行目标测距，当合成孔径长度较大时，获得聚焦清晰的像是困难的，桶型失真等因素使得通用的聚焦校正十分复杂，这就需要在合成孔径聚焦不好的条件下测量目标距离，这正是双目合成孔径测距的意义。双目合成孔径左右面阵在设计上具有一致性，使得合成孔径误差也具有一致性，在误差分析中一致误差又称为共模误差，差分处理可抑制共模误差。系统设计使得双目合成孔径聚焦共模误差占有较大比重，为抑制误差测距创造了条件。目标距离的检测受目标像邻域大小影响，因聚散焦效应，不在聚焦距离的目标信息熵被衰减，图像信息熵更容易来自同一距离的目标，因此合成孔径叠加像中目标邻域可以相对大，以提高测距信息来源。Large-aperture lenses only focus on objects at certain distances, while defocusing objects at other distances. The target information is measured in the form of image entropy. The focused image maintains the information entropy of the target image, while the defocus effect attenuates the target information. Therefore, the target distance can be judged by detecting the retention of the target image information entropy. The retention of target image information entropy can theoretically be measured by the amount of compatible information between left and right synthetic aperture images, which is the basic method of binocular synthetic aperture focusing distance measurement. The amount of compatible information between two images refers to the amount of the same part of the information of the two images, and its specific measurement can take various reasonable forms. Preferably, for grayscale images, the measurement of the amount of compatible information The form is the product of the information entropy of the two images and the similarity between the two images, or the product of the average information entropy of the two images and the similarity between the two images, or the mutual entropy of the two images; for color images, the The measurement form of the above-mentioned consistent information content is the sum or mean value of the consistent information content between the sub-images of each channel in the RGB color space of the two images obtained by using the consistent information content measurement form of the grayscale image. Of course, other measurement forms can also be used. For example, the information entropy in the above measurement form can be replaced by the sharpness or gradient or contrast of two images, and the measurement of compatible information can also be realized; It is a measurement form with the same change trend of content information. Studholme proposed normalized mutual information (NormalizedMutualInformation, NMI): where (x, y) is the pixel coordinates, H(R) and H(F) are the information entropy of the reference image and the synthesized image in the neighborhood image blocks at the coordinates (x, y), and the joint entropy of H(R, F) , the most commonly used histogram method for calculation, which highlights the frequency instead of probability for density estimation from an intuitive point of view. In this specific implementation mode, the histogram method is used to calculate the mutual information entropy to measure the amount of compatible information. Using the monocular synthetic aperture focusing image for target ranging, when the synthetic aperture length is large, it is difficult to obtain a clear focus image, barrel distortion and other factors make the general focus correction very complicated, which requires the synthetic aperture focusing not Measuring the target distance under good conditions is exactly the meaning of binocular synthetic aperture ranging. The design of the binocular synthetic aperture left and right arrays is consistent, which makes the synthetic aperture error also consistent. In the error analysis, the consistent error is also called the common mode error, and the differential processing can suppress the common mode error. The system design makes the binocular synthetic aperture focusing common mode error account for a large proportion, creating conditions for suppressing error ranging. The detection of the target distance is affected by the size of the target image neighborhood. Due to the defocusing effect, the information entropy of the target not at the focus distance is attenuated, and the image information entropy is more likely to come from the target at the same distance. Therefore, the target neighborhood in the synthetic aperture superimposed image can be relatively large to improve ranging information sources.
下面以一个具体实施例来对本发明技术方案进行进一步详细说明。本发明基于双目合成孔径聚焦图像的目标测距方法，如图2所示，具体包括以下步骤：The technical solution of the present invention will be further described in detail with a specific embodiment below. The present invention is based on the target distance measuring method of binocular synthetic aperture focusing image, as shown in Figure 2, specifically comprises the following steps:
步骤1、从两个结构相同的面阵处，以所述两个面阵的对称中心作为基准机位，对目标所在场景分别进行光学合成孔径成像，分别得到一组对应不同聚焦距离的合成孔径图像。Step 1. From two area arrays with the same structure, take the symmetry center of the two area arrays as the reference camera position, perform optical synthetic aperture imaging on the scene where the target is located, and obtain a set of synthetic apertures corresponding to different focusing distances image.
从两个面阵处分别进行光学合成孔径成像，即可得到双目合成孔径图像，即两组对应不同聚焦距离的合成孔径图像。合成孔径聚焦成像要求对特定距离清晰聚焦，其他距离处的目标像散焦为均匀平滑的像斑，为避免桶型失真校正，以实现实时测量，本发明优选采用两个结构相同的圆形面阵；所述光学合成孔径成像可通过密布在圆形面阵中的小孔成像模型摄像机阵列同时拍摄实现，或者通过单个小孔成像模型摄像机在圆形面阵中一系列密集排布的机位分时拍摄实现。如图3所示，左右面阵密布在两个大圆内，使光瞳成圆形，以减小合成孔径长度。当桶型失真不可避免时，可采用插值、拆分摄像机图像画面以减小视场锥角等方法校正桶型失真，这一过程可结合高分辨率重建同步进行。摄像机阵的各摄像机成像符合小孔成像模型，双目合成孔径成像的两面阵对称中心为基准机位。合成孔径聚焦立体视觉要求获得同一目标不同机位的多个图像，对光照、摄像机差异、噪声等有较强的适应能力。图像序列可以来自摄像机阵列，也可以是同一摄像机移位分时拍摄。移动视频图像序列可以组合成双目合成孔径聚焦成像图像序列，实际上航拍图像序列处理，也可采用双目合成孔径原理获得深度信息。此外，需要特别指出的是，一些研究者提出将图像超分辨率重建技术与光学合成孔径成像技术相结合，以突破图像采集设备固有分辨率限制，得到更高分辨率合成孔径图像。本步骤中也可使用现有或将有的各种基于超分辨率复原的光学合成孔径成像方法对目标所在场景分别进行光学合成孔径成像，所获得的超分辨率合成孔径聚焦图像可进一步提高目标测距的精度。The binocular synthetic aperture image can be obtained by performing optical synthetic aperture imaging from the two arrays, that is, two sets of synthetic aperture images corresponding to different focusing distances. Synthetic aperture focusing imaging requires clear focus at a specific distance, and the target astigmatism at other distances is a uniform and smooth image spot. In order to avoid barrel distortion correction and realize real-time measurement, the present invention preferably uses two circular surfaces with the same structure array; the optical synthetic aperture imaging can be realized by simultaneous shooting of a small hole imaging model camera array densely distributed in a circular array, or a series of densely arranged camera positions in a circular array through a single small hole imaging model camera Time-sharing shooting is realized. As shown in Figure 3, the left and right arrays are densely packed in two large circles, so that the pupil becomes circular to reduce the length of the synthetic aperture. When barrel distortion is unavoidable, methods such as interpolation and splitting camera images to reduce the cone angle of the field of view can be used to correct barrel distortion. This process can be combined with high-resolution reconstruction simultaneously. The imaging of each camera in the camera array conforms to the pinhole imaging model, and the symmetry center of the binocular synthetic aperture imaging on both sides of the array is the reference camera position. Synthetic aperture focusing stereo vision requires multiple images of the same target from different camera positions, and has strong adaptability to illumination, camera differences, noise, etc. The image sequence can come from a camera array, or it can be time-shared shooting with the same camera shifted. The moving video image sequence can be combined into a binocular synthetic aperture focusing imaging image sequence. In fact, the aerial image sequence processing can also use the binocular synthetic aperture principle to obtain depth information. In addition, it needs to be pointed out that some researchers proposed to combine image super-resolution reconstruction technology with optical synthetic aperture imaging technology to break through the inherent resolution limitation of image acquisition equipment and obtain higher resolution synthetic aperture images. In this step, various existing or future optical synthetic aperture imaging methods based on super-resolution restoration can also be used to perform optical synthetic aperture imaging on the scene where the target is located, and the obtained super-resolution synthetic aperture focused image can further improve the target image. The accuracy of ranging.
下面以垂直目标匀速移动拍摄的视频图像序列为例来说明获得双目合成孔径图像的过程：The following takes the video image sequence shot by a vertical target moving at a constant speed as an example to illustrate the process of obtaining a binocular synthetic aperture image:
步骤1-1、采集双目摄像机阵列图像：Step 1-1. Collect binocular camera array images:
图4为匀速移动机位拍摄的60幅图像序列，拍摄用摄像机采用超景深的小孔镜头，图像序列是通过单摄像机分时移动采集得到(当然，也可以采用摄相机阵列同时拍摄)。将图像序列按机位接近原则分为两组，每组30幅组成左右阵列。Figure 4 is a sequence of 60 images shot by moving cameras at a constant speed. The camera used for shooting uses a small hole lens with super depth of field. Divide the image sequence into two groups according to the principle of camera position proximity, and each group has 30 images to form a left and right array.
步骤1-2、计算各聚焦距离下图像与基准机位间的视差位移，左、右阵列各聚焦距离下图像分别校正视差后叠加。Step 1-2. Calculate the parallax displacement between the image at each focus distance and the reference camera position, and superimpose the images at each focus distance of the left and right arrays after correcting the parallax.
根据摄像机参数、目标距离和图3几何关系可以得出摄像机机位与基准机位的视差，基准机位位于图3画面的对称中心，即图中箭头指示处。在基准机位不一定要采集图像，这并不影响双目阵图像以此机位为参照校正视差。将校正视差后的左面阵各图像叠加为左目合成孔径图像，将校正视差后的右面阵各图像叠加为右目合成孔径图像，校正视差移位是亚像素级，可以使合成孔径像聚焦在任意距离处，即成像聚焦面可以任意移动。各组摄像机阵最长直径长度为合成孔径长度。这样，对于左、右阵列，既可分别得到一组对应不同聚焦距离的合成孔径图像。According to the camera parameters, target distance and the geometric relationship in Figure 3, the parallax between the camera position and the reference position can be obtained. The reference position is located at the symmetrical center of the picture in Figure 3, which is indicated by the arrow in the figure. It is not necessary to collect images at the reference camera position, which does not affect the parallax correction of binocular array images using this camera position as a reference. The images of the left array after parallax correction are superimposed into the left-eye synthetic aperture image, and the images of the right array after parallax correction are superimposed into the right-eye synthetic aperture image. The corrected parallax shift is at the sub-pixel level, and the synthetic aperture image can be focused at any distance , that is, the imaging focal plane can be moved arbitrarily. The length of the longest diameter of each group of camera arrays is the length of the synthetic aperture. In this way, for the left and right arrays, a group of synthetic aperture images corresponding to different focusing distances can be respectively obtained.
图5显示了所得到的部分合成孔径图像，其中(a)为左目聚焦在车辆前沿处，(b)为左目聚焦在相对远景处，(c)为右目聚焦在车辆前沿处，(d)为右目聚焦在相对远景处。由于合成孔径直径比光学镜头大得多，所以微小的目标距离差异，也能造成明显的聚/散焦效应。由图5可见，特定距离处目标左右阵聚焦像位置重合，其他距离处不但目标成散焦像，而且位置发生偏移。图像序列合成像对单个图像光照、摄像机抖动、噪声等差异有明显抑制能力。合成误差使得合成像与单幅像相比，清晰度明显下降。而双目合成孔径聚焦测距可抑制共模合成误差，因此允许合成像因误差造成聚焦像清晰度有所下降。Figure 5 shows part of the obtained synthetic aperture images, where (a) is the left eye focusing on the front of the vehicle, (b) is the left eye focusing on the relative distance, (c) is the right eye focusing on the front of the vehicle, (d) is The right eye focuses on the relative distance. Since the diameter of the synthetic aperture is much larger than that of the optical lens, a slight difference in the target distance can also cause obvious focusing/defocusing effects. It can be seen from Figure 5 that the positions of the left and right focused images of the target at a specific distance coincide, while at other distances the target not only forms a defocused image, but also shifts its position. Image sequence composite imaging has obvious ability to suppress differences in single image illumination, camera shake, noise, etc. Composite errors make composite images significantly less clear than single images. The binocular synthetic aperture focusing distance measurement can suppress the common-mode synthetic error, so the clarity of the focused image is allowed to decrease due to the error of the synthetic image.
步骤2、对每一个聚焦距离，将其所对应的从两个面阵处获得的两幅合成孔径图像作为一对，然后对每一对合成孔径图像进行以下处理：以其中一幅图像中的目标图像作为参考图像，将另一幅图像中与参考图像相对应的区域作为匹配位置图像，并将另一幅图像中与匹配位置图像间的像差矢量的模在(0,d]范围内变化的一系列区域作为非匹配位置图像，d为预设的最大像差；计算参考图像与匹配位置图像之间的相容信息量R0以及参考图像与各非匹配位置图像之间的相容信息量，并将所计算出的参考图像与各非匹配位置图像之间的相容信息量中大于R0的值统一修改为R0，然后计算修改后的参考图像与各非匹配位置图像之间的相容信息量的均值；最后以R0与所述均值的差值作为该聚焦距离的检测量。Step 2. For each focusing distance, take the corresponding two synthetic aperture images obtained from the two area arrays as a pair, and then perform the following processing on each pair of synthetic aperture images: The target image is used as a reference image, and the area corresponding to the reference image in the other image is used as the matching position image, and the modulus of the aberration vector between the other image and the matching position image is in the range of (0, d] A series of changing areas are used as non-matching position images, d is the preset maximum aberration; calculate the compatible information amount R0 between the reference image and the matching position image and the compatibility between the reference image and each non-matching position image Information amount, and uniformly modify the value greater than R0 in the calculated compatible information amount between the reference image and each non-matching position image to R0 , and then calculate the difference between the modified reference image and each non-matching position image The average value of the consistent information amount between; finally, the difference between R0 and the average value is used as the detection amount of the focus distance.
对于某一聚焦距离z，判断目标像是否聚焦，不但要考察左、右目合成孔径图像中目标像间的相容信息量是否达到最大，还要考察左、右目标像位置是否在理想位置重合，设定聚焦距离处目标左、右目标像位置重合，其他聚焦距离处目标成散焦像，位置发生偏移。因此对每个聚焦距离的左右目标像，需搜索计算一定像差距离范围内(由于本实施例中采用线阵合成孔径成像，仅需在[-d,d]的像差范围内搜索，d为预设的最大像差，搜索时最好在该范围内逐像素变化；如果采用优选的圆形面阵，则需要在像差矢量的模为[0,d]范围内搜索)目标像间的相容信息量，若相容信息量最大值出现在非理想位，则表明该区域散焦像干扰测距，需抑制此干扰。图6显示了不同目标的双目目标图像相容信息量随聚焦距离和视差的变化分布情况，(a)中的十字线交点为需测距点，处于建筑边沿线上，它的图像熵较大；(b)为该点邻域左、右目合成孔径图像间的相容信息量Rzd(x,y)随聚焦距离z和匹配像差变化的二维分布图，图中亮度与相容信息量大小正相关；图(b)的水平向为聚焦距离参数z，从左向右，聚焦距离参数从15米变化到5米，图(b)垂直向为像差，从下向上像差参数从-20像素变化到+20像素。由图6中的(b)可见相容信息量最大正是出现在像差0处，此处距离参数正是目标的真实距离，参见图7中的(a)。当需测距点换为图6的图(c)中的十字线交点处，它的图像熵较小，而附近的建筑边沿图像熵较大，它会干扰测距，图6中的图(d)为该点邻域左、右目合成孔径图像间的相容信息量随聚焦距离和匹配像差变化的二维分布图，可见相容信息量最大值出现在像差-13像素处，此处的距离参数并不是目标的真实距离，参见图7中的图(b)。为此，本发明对所计算出来的参考图像与各非匹配位置图像之间的相容信息量进行了削峰处理，即将所计算出的参考图像与各非匹配位置图像之间的相容信息量中大于Rz0(x,y)的Rzd(x,y)统一降为Rz0(x,y)，在一定程度上抑制了其它目标散焦像干扰，经此处理，目标点邻域左、右目合成孔径图像间的相容信息量随聚焦距离和匹配像差变化的二维分布如图6中的图(e)所示。For a certain focusing distance z, to judge whether the target image is in focus, it is not only necessary to examine whether the amount of compatible information between the target images in the left and right synthetic aperture images reaches the maximum, but also to examine whether the positions of the left and right target images overlap at the ideal position, The positions of the left and right target images at the set focus distance coincide, and the targets at other focus distances become defocused images, and their positions are shifted. Therefore, for the left and right target images of each focusing distance, it is necessary to search and calculate within a certain range of aberration distances (because linear array synthetic aperture imaging is used in this embodiment, it is only necessary to search within the range of aberrations of [-d, d], d is the preset maximum aberration, it is better to change pixel by pixel within this range when searching; if the preferred circular area array is used, it is necessary to search within the range of the modulus of the aberration vector [0,d]) between the target images The amount of compatible information, if the maximum value of the amount of compatible information appears in the non-ideal position, it indicates that the defocused image in this area interferes with ranging, and this interference needs to be suppressed. Figure 6 shows the distribution of compatible information of binocular target images of different targets with the change of focusing distance and parallax. The intersection point of the cross line in (a) is the point to be measured, which is on the edge of the building, and its image entropy is relatively high. (b) is the two-dimensional distribution diagram of the compatible information amount Rzd (x, y) between the left and right synthetic aperture images in the neighborhood of this point as the change of focusing distance z and matching aberration, in which the brightness and compatible The amount of information is positively correlated; the horizontal direction of the picture (b) is the focusing distance parameter z, from left to right, the focusing distance parameter changes from 15 meters to 5 meters, the vertical direction of the picture (b) is the aberration, and the aberration is from the bottom to the top The parameter varies from -20 px to +20 px. It can be seen from (b) in Figure 6 that the maximum amount of compatible information appears at the aberration 0, where the distance parameter is the real distance of the target, see (a) in Figure 7. When the distance-measuring point is replaced by the intersection point of the crosshairs in Figure (c) of Figure 6, its image entropy is small, while the image entropy of nearby building edges is relatively large, which will interfere with distance measurement, and the figure in Figure 6 ( d) It is a two-dimensional distribution map of the compatible information content between the left and right synthetic aperture images in the neighborhood of the point as a function of the focusing distance and matching aberration. It can be seen that the maximum compatible information content appears at the aberration -13 pixels, here The distance parameter at is not the real distance of the target, see graph (b) in Figure 7. For this reason, the present invention performs peak-shaving processing on the compatible information between the calculated reference image and each non-matching position image, that is, the compatible information between the calculated reference image and each non-matching position image Rzd (x, y) that is greater than Rz0 (x, y) in the quantity is uniformly reduced to Rz0 (x, y), which suppresses the interference of other target defocused images to a certain extent. After this processing, the neighborhood of the target point The two-dimensional distribution of the amount of compatible information between the left and right synthetic aperture images as a function of focusing distance and matching aberration is shown in Figure 6 (e).
在经过削峰处理从而抑制散焦像信号基础上，计算削峰处理后的参考图像与各非匹配位置图像之间的相容信息量的均值得到一平稳的相容信息量参照值，将参考图像与匹配位置图像之间的相容信息量与该参照值的差作为检测量，显示了二目视觉像匹配与非匹配的差异度，对聚焦像此差异度大，对散焦像此差异度小。此检测量不受左右合成孔径聚焦共模误差影响，此检测量最大值对应着目标距离。对于图6的图(a)中的目标，检测量随距离变化曲线如图7中的图(a)所示，可见检测量最大值处于13米处。对于图6的图(c)中的目标，其参考图像与匹配位置图像之间的相容信息量随距离变化曲线如图7中的图(b)所示，其最大值对应错误距离参数，而以双目匹配位置相容信息量与非匹配位置相容信息量均值之差作为检测量，可正确检测出目标处于正确距离13米处，如图7中的图(c)所示。因此本发明所提出的检测量可作为目标测距的准确衡量指标。On the basis of peak clipping processing to suppress the defocused image signal, calculate the average value of the compatible information between the reference image after peak clipping processing and each non-matching position image Obtain a stable reference value of consistent information, and the difference between the consistent information between the reference image and the matching position image and the reference value As the detection quantity, it shows the degree of difference between matching and non-matching of binocular vision images, the degree of difference is large for in-focus images, and the degree of difference for defocused images is small. This detection amount is not affected by the common mode error of the left and right synthetic aperture focusing, and the maximum value of this detection amount corresponds to the target distance. For the target in Figure (a) of Figure 6, the detection amount varies with distance as shown in Figure (a) in Figure 7, and it can be seen that the maximum detection amount is at 13 meters. For the target in Figure (c) of Figure 6, the variation curve of the amount of compatible information between its reference image and the matching position image with distance is shown in Figure (b) in Figure 7, and its maximum value corresponds to the error distance parameter, The difference between the binocular matching position compatible information volume and the non-matching position compatible information volume mean As the detection quantity, it can be correctly detected that the target is at the correct distance of 13 meters, as shown in graph (c) in FIG. 7 . Therefore, the detection quantity proposed by the present invention can be used as an accurate measurement index for target distance measurement.
此外，为了防止目标像信息不足所导致的测距错误，本发明对最大相容信息量设一门限值，小于门限表明该目标像缺乏灰度纹理等信息，不能测距；大于门限的测距为有效测距。具体如下：In addition, in order to prevent ranging errors caused by insufficient target image information, the present invention sets a threshold value for the maximum compatible information amount, which indicates that the target image lacks information such as gray scale texture and cannot measure distance if it is less than the threshold value; The distance is the effective distance measurement. details as follows:
步骤3、找出检测量最大的聚焦距离，并判断该聚焦距离所对应的R0是否大于预设的相容信息量门限值，如是，则以该聚焦距离作为目标距离。Step 3. Find out the focus distance with the largest detection amount, and judge whether the R0 corresponding to the focus distance is greater than the preset compatible information amount threshold, and if so, use the focus distance as the target distance.
图8显示了最终获得的目标深度图，图中左侧为亮度距离比照条，深黑处为不可测距点。加上街道模型和车辆模型先验信息，可分析出道路上各车道占用情况，规划出自动驾驶车辆可用车道，为自动驾驶车辆导航。完整的立体视觉还包括三维建模的平滑和精细化工作，但在自动驾驶车辆导航中可省略。Figure 8 shows the final target depth map. The left side of the figure is the brightness distance comparison bar, and the dark place is the unmeasured point. In addition to the prior information of the street model and vehicle model, the occupancy of each lane on the road can be analyzed, and the available lanes for autonomous vehicles can be planned to guide autonomous vehicles. Full stereo vision also includes the smoothing and refinement of 3D modeling, which can be omitted in autonomous vehicle navigation.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510732130.9ACN105258673B (en) | 2015-11-02 | 2015-11-02 | A kind of target ranging method based on binocular synthetic aperture focusing image, device |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510732130.9ACN105258673B (en) | 2015-11-02 | 2015-11-02 | A kind of target ranging method based on binocular synthetic aperture focusing image, device |
| Publication Number | Publication Date |
|---|---|
| CN105258673Atrue CN105258673A (en) | 2016-01-20 |
| CN105258673B CN105258673B (en) | 2017-05-31 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510732130.9AActiveCN105258673B (en) | 2015-11-02 | 2015-11-02 | A kind of target ranging method based on binocular synthetic aperture focusing image, device |
| Country | Link |
|---|---|
| CN (1) | CN105258673B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107716468A (en)* | 2017-10-31 | 2018-02-23 | 广东工业大学 | A kind of laser auto focusing method, system, device and readable storage medium storing program for executing |
| CN107820093A (en)* | 2017-11-15 | 2018-03-20 | 深圳大学 | Information detecting method, device and receiving device based on packet energy differences |
| CN108226901A (en)* | 2018-02-06 | 2018-06-29 | 北京万集科技股份有限公司 | Laser radar optical system |
| CN110749897A (en)* | 2018-07-23 | 2020-02-04 | Ibeo汽车系统有限公司 | Method and device for optical distance measurement |
| CN110910436A (en)* | 2019-10-30 | 2020-03-24 | 深圳供电局有限公司 | Ranging method, device, device and medium based on image information enhancement technology |
| CN113391644A (en)* | 2021-06-11 | 2021-09-14 | 中新国际联合研究院 | Unmanned aerial vehicle shooting distance semi-automatic optimization method based on image information entropy |
| CN116228547A (en)* | 2023-05-08 | 2023-06-06 | 天府兴隆湖实验室 | Simulation algorithm of synthetic aperture imaging process |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102867304A (en)* | 2012-09-04 | 2013-01-09 | 南京航空航天大学 | Method for establishing relation between scene stereoscopic depth and vision difference in binocular stereoscopic vision system |
| CN103026171A (en)* | 2011-05-27 | 2013-04-03 | 松下电器产业株式会社 | Image processing apparatus and image processing method |
| CN103033166A (en)* | 2012-12-13 | 2013-04-10 | 南京航空航天大学 | Target ranging method based on synthetic aperture focused images |
| US20140071427A1 (en)* | 2012-09-07 | 2014-03-13 | Apple Inc. | Imaging range finder fabrication |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103026171A (en)* | 2011-05-27 | 2013-04-03 | 松下电器产业株式会社 | Image processing apparatus and image processing method |
| CN102867304A (en)* | 2012-09-04 | 2013-01-09 | 南京航空航天大学 | Method for establishing relation between scene stereoscopic depth and vision difference in binocular stereoscopic vision system |
| US20140071427A1 (en)* | 2012-09-07 | 2014-03-13 | Apple Inc. | Imaging range finder fabrication |
| CN103033166A (en)* | 2012-12-13 | 2013-04-10 | 南京航空航天大学 | Target ranging method based on synthetic aperture focused images |
| Title |
|---|
| 周致远: "接近段目标双目视觉测量中的精度分析与像机标定方法研究", 《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 |
|---|---|---|---|---|
| CN107716468A (en)* | 2017-10-31 | 2018-02-23 | 广东工业大学 | A kind of laser auto focusing method, system, device and readable storage medium storing program for executing |
| CN107820093A (en)* | 2017-11-15 | 2018-03-20 | 深圳大学 | Information detecting method, device and receiving device based on packet energy differences |
| CN107820093B (en)* | 2017-11-15 | 2019-09-03 | 深圳大学 | Information detection method, device and receiving device based on packet energy difference |
| CN108226901A (en)* | 2018-02-06 | 2018-06-29 | 北京万集科技股份有限公司 | Laser radar optical system |
| CN108226901B (en)* | 2018-02-06 | 2024-03-01 | 武汉万集光电技术有限公司 | Laser radar optical system |
| CN110749897B (en)* | 2018-07-23 | 2023-06-23 | 微视公司 | Method and device for optical distance measurement |
| CN110749897A (en)* | 2018-07-23 | 2020-02-04 | Ibeo汽车系统有限公司 | Method and device for optical distance measurement |
| CN110910436A (en)* | 2019-10-30 | 2020-03-24 | 深圳供电局有限公司 | Ranging method, device, device and medium based on image information enhancement technology |
| CN110910436B (en)* | 2019-10-30 | 2022-10-28 | 深圳供电局有限公司 | Ranging method, device, device and medium based on image information enhancement technology |
| CN113391644A (en)* | 2021-06-11 | 2021-09-14 | 中新国际联合研究院 | Unmanned aerial vehicle shooting distance semi-automatic optimization method based on image information entropy |
| CN113391644B (en)* | 2021-06-11 | 2022-06-17 | 中新国际联合研究院 | Unmanned aerial vehicle shooting distance semi-automatic optimization method based on image information entropy |
| CN116228547A (en)* | 2023-05-08 | 2023-06-06 | 天府兴隆湖实验室 | Simulation algorithm of synthetic aperture imaging process |
| CN116228547B (en)* | 2023-05-08 | 2023-07-07 | 天府兴隆湖实验室 | Simulation method of synthetic aperture imaging process |
| Publication number | Publication date |
|---|---|
| CN105258673B (en) | 2017-05-31 |
| Publication | Publication Date | Title |
|---|---|---|
| CN105258673B (en) | A kind of target ranging method based on binocular synthetic aperture focusing image, device | |
| CN110288642B (en) | Three-dimensional object rapid reconstruction method based on camera array | |
| CN103279982B (en) | The speckle three-dimensional rebuilding method of the quick high depth resolution of robust | |
| CN111563921B (en) | An underwater point cloud acquisition method based on binocular camera | |
| EP3869797B1 (en) | Method for depth detection in images captured using array cameras | |
| EP2568253B1 (en) | Structured-light measuring method and system | |
| CN107977997A (en) | A kind of Camera Self-Calibration method of combination laser radar three dimensional point cloud | |
| CN109285189B (en) | A fast calculation method of linear trajectory without binocular synchronization | |
| CN106225676B (en) | Method for three-dimensional measurement, apparatus and system | |
| CN102831601A (en) | Three-dimensional matching method based on union similarity measure and self-adaptive support weighting | |
| CN112465912A (en) | Three-dimensional camera calibration method and device | |
| CN105160680A (en) | Design method of camera with no interference depth based on structured light | |
| KR101709317B1 (en) | Method for calculating an object's coordinates in an image using single camera and gps | |
| CN103971378A (en) | Three-dimensional reconstruction method of panoramic image in mixed vision system | |
| CN106296811A (en) | A kind of object three-dimensional reconstruction method based on single light-field camera | |
| CN109712232B (en) | Object surface contour three-dimensional imaging method based on light field | |
| CN103299343A (en) | Range image pixel matching method | |
| CN103793911A (en) | Scene depth obtaining method based on integration image technology | |
| CN103033166B (en) | Target ranging method based on synthetic aperture focused images | |
| KR20060063558A (en) | Depth Information Based Stereo / Multipoint Image Matching Apparatus and Method | |
| CN103776419A (en) | Binocular-vision distance measurement method capable of widening measurement range | |
| CN114359406A (en) | Calibration of auto-focusing binocular camera, 3D vision and depth point cloud calculation method | |
| Patel et al. | Distance measurement system using binocular stereo vision approach | |
| CN106203429A (en) | Based on the shelter target detection method under binocular stereo vision complex background | |
| CN108924408A (en) | A kind of Depth Imaging method and system |
| 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 |