Research On Stereo Matching Of Binocular Medical Endoscope

Endoscopes play a key role in medical diagnosis and surgical operations.Traditional two-dimensional endoscopes cannot provide the exact location of tissues and organs,which is not conducive to the diagnosis and treatment of diseases by doctors.To this end,this article explores the binocular stereoscopic medical endoscope,focusing on the preprocessing of the endoscopic image and the stereo matching technology.In the preprocessing process of the endoscopic image,the Bouguet method is used to complete the stereo correction of the original image pair;then the original RGB endoscopic image is converted into different color spaces to study the effect of corresponding histogram equalization.The results show that the use of The final RGB image obtained by the histogram equalization in the YUV color space satisfies the human eye observation requirements,and the histogram statistical results of each channel in the resulting image are also more uniform.In terms of feature point extraction and matching,the ratio method and the RANSAC method are used to screen the initial matching results.The results show that the SURF matching algorithm can meet the requirements of the accuracy and operating efficiency of the endoscopic position estimation.Regarding the dense reconstruction of endoscopic images,in view of the particularity of the endoscopic environment,the traditional guided filtering algorithm is improved,and a stereo matching algorithm based on guided filtering and disparity map fusion is proposed.The algorithm divides the original image into two regions by constructing the shortest arm length image and threshold segmentation for targeted processing,and then uses the disparity map fusion method to fuse the results of different regions.The proposed algorithm is verified experimentally on Middlebury standard test data set images.The results show that the average mismatch rate of the improved algorithm on 6 sets of weak texture images is only 9.67%.It shows that the improved algorithm has better performance in the weak texture area of the image,can effectively improve the stereo matching accuracy of medical endoscopic images,and is helpful for the development of dense reconstruction of endoscopic images.