The high definition miniature electronic endoscopic system of binocular visionTechnical field
The present invention relates to a kind of technology of medical field, peeped in the high definition miniature electronic of specifically a kind of binocular visionMirror system.
Background technique
Only one imaging sensor of cannula type endoscopic system, can only obtain flat image, can only in flat imageThe relative positional relationship between image is obtained, the accurate size information of image can not be obtained, and in clinical application, it is peeped by interiorSem observation to image quantitatively to obtain its accurate size information extremely important, such as determine wound size size, lesion area, it is existingSome cannula type endoscopes are unable to complete above-mentioned requirements.
Summary of the invention
The present invention is proposed for the defects of prior art size is big, power consumption is high, structure is complicated and image quality is affectedA kind of high definition miniature electronic endoscopic system of binocular vision, can obtain high-definition image in real time, will acquire by dsp processorBinocular image merged, realize the three-dimensional reconstruction of shooting image, obtain the information such as size, size, depth of shooting image,Accurate quantitative analysis and medical endoscope image high-definition are provided for surgical operation, improves the standard of surgical operation diagnosisTrue rate.
The present invention is achieved by the following technical solutions:
The present invention includes: two groups of mutually isostructural image acquisition units, the number letter being sequentially connected with image acquisition unitsNumber processing module and binocular vision construct module, in which: image acquisition units include sequentially connected optical fiber, camera lens, miniature heightClear imaging sensor and image capture module, optical fiber provide illumination, Image Acquisition mould by light source-guide to camera lens front end, for camera lensThe signal output end of block by LVDS conducting wire be connected with LVDS interface simultaneously output digital image signal to digital signal processing module,Digital signal processing module carries out geometric dimension measurement to data image signal and exports geological information to binocular vision building mouldBlock realizes binocular vision after binocular vision building module carries out match point reparation.
The image acquisition units are set to inside cylinder housing, in which: two optical fiber and two camera lenses are distributed inOn the cross section of cylinder housing.
The geometric dimension measurement refers to: digital signal processing module carries out data image signal according to calibrating parametersAfter distortion correction and polar curve are handed over just, for the certain point in binocular image in the image of side, its local feature information structuring is utilizedLocal feature description's, it is special by same structure method construct part for all the points gone together in the figure of the other side with the figure pointSign description, calculates the Euclidean distance between the Feature Descriptor of two sides picture point, selected distance minimum value is other side figureIn match point, obtain three-dimensional coordinate of the left and right match points in world coordinate system using triangle geometric method, and by multipleIt is mutually matched a little to carry out length, width, depth, area and cubing and obtain geological information.
The match point reparation refers to: causing match point for mistake when the part in the figure of side is blocked in the figure of the other sideWhen match point, Mismatching point is rejected by consistency check, and for the point of non-successful match, using with its most neighborThe parallax value successfully put is carried out the calculating of three-dimensional coordinate using principle of triangulation, obtains three of all the points on the image of two sidesDimension coordinate is to realize binocular vision.
Technical effect
Compared with prior art, the present invention can obtain the three of the geometry information for shooting image and shooting image in real timeImage is tieed up, provides the geometry information of lesion accurate quantitative analysis, compared with prior art, this hair for the diagnosis decision of clinicianIt is bright that the flat image information of shooting image can not only be provided for clinician, additionally it is possible to which that the geometric dimension letter of accurate quantitative analysis is providedBreath and three-dimensional image information provide more acurrate and quantitative information to the correct diagnosis of clinician and decision.
Detailed description of the invention
Fig. 1 is structure chart of the invention；
The top view of the end face of the present invention Fig. 2；
Fig. 3 is left images captured by the present invention；
Fig. 4 is the image of three-dimensional reconstruction of the present invention；
Wherein: camera lens 1, Miniature high-definition imaging sensor 2, image capture module 3, power supply line 4, LVDS conducting wire 5, cylindrical bodyShell 6, LVDS interface 7, digital signal processing module 8, USB3.0 interface 9, computer 10, image processing module 11, optical fiber 12.
Specific embodiment
As shown in Figure 1, the present embodiment include: two groups of mutually isostructural image acquisition units, with image acquisition units successivelyConnected digital signal processing module 8, the computer 10 with image processing module 11, in which: image acquisition units include according toOptical fiber 12, camera lens 1, Miniature high-definition imaging sensor 2 and the image capture module 3 of secondary connection, optical fiber 12 is by light source-guide to mirrorFirst 1 front end provides illumination for camera lens 1, and the signal output end of image capture module 3 is connected by LVDS conducting wire 5 with LVDS interface 7And output digital image signal is to digital signal processing module 8.
The digital signal processing module 8 is connected by USB3.0 interface 9 with computer 10.
The image acquisition units are set to inside cylinder housing 6, in which: two optical fiber 12 and two camera lenses 1 are equalIt is distributed on the cross section of cylinder housing 6.
The mutually isostructural image acquisition units parameter of described two groups is identical, by identical camera lens, LVDS conducting wire andLVDS interface is sent to dsp processor, and two sets of high-definition images are sent to computer, computer by USB3.0 by dsp processorTwo sets of images are carried out three-dimensional reconstruction by upper image processing module 11, obtain the 3D rendering of shooting object.
As shown in figure 3, the image acquisition units carry out geometric dimension measurement to shooting image by following steps:
The first step uses 13 rows, 8 column, gridiron pattern of the cell having a size of 1.5mm square as calibration chessboard, inside peepsMirror with different view takes pictures 25 to chessboard with distance, carries out parameter to two Miniature high-definition cameras using Zhang Zhengyou calibration methodCalibration；
Second step, the parameter obtained using calibration carries out distortion correction to image and polar curve is handed over just；
Third step, for the certain point in left image, using its local feature information structuring local feature description, forThe all the points gone together in right figure with left figure point pass through same structure method construct local feature description；
4th step, computer calculate third step in left images point Feature Descriptor between Euclidean distance, choose away fromFrom the match point that minimum value is in right figure；
5th step is obtained three-dimensional coordinate of the left and right match points in world coordinate system using triangle geometric method, and passed throughIt is multiple to be mutually matched a little to carry out length, width, depth, area and cubing.
As shown in figure 4, the image processing module 11 carries out match point reparation to shooting image by following steps:
The first step, according to from left to right, sequence from top to bottom successively traverses the every bit in left image, and in right figureThe point to match is searched in corresponding row, the three-dimensional coordinate of the point is obtained according to the parallax of match point；
Due to visual angle, the part in left figure may will appear the case where blocking in right figure, the matching obtained at this timePoint is Mismatching point, is in the case rejected Mismatching point using left and right consistency check；
Second step, the point for non-successful match utilize triangle using the parallax value successfully put with its most neighborMeasuring principle carries out the calculating of three-dimensional coordinate, finally obtains the three-dimensional coordinate of all the points on image and is achieved binocular vision.
By above method, the present invention can obtain the geometry information of shooting image in real time and shoot the three-dimensional of imageImage provides the geometry information of lesion accurate quantitative analysis for the diagnosis decision of clinician, compared with prior art, the present inventionThe flat image information of shooting image can not only be provided for clinician, additionally it is possible to provide the geometry information of accurate quantitative analysisAnd three-dimensional image information, accurate and quantitative information is provided to the correct diagnosis of clinician and decision.
Above-mentioned specific implementation can by those skilled in the art under the premise of without departing substantially from the principle of the invention and objective with differenceMode carry out local directed complete set to it, protection scope of the present invention is subject to claims and not by above-mentioned specific implementation instituteLimit, each implementation within its scope is by the constraint of the present invention.