The Motion Trajectory Detection For Specific Objective Based On Computer Vision

In the field of computer science and artificial intelligence, computer vision is one of the most popular research in this year. The measuring method based on computer vision can break through the conditional restriction of the conventional measurement. The positioning and tracking of surgical instruments is an integral part of Computed Assisted Surgery, apply the trajectory detection based on computer vision to it, assist the surgeon to complete the difficult operation, that is of vital importance to the progress of medical technology.The main task of this paper is to detect the track of specific objective based on computer vision, mainly divided into four aspects: camera calibration, image processing,stereo matching and object tracking. Considering that the application environment requires high-precision location, strong real-time and well stability, The calibration method of the camera calibration selected Zhang's plane method, it is adopted to flexibly an quickly get the exterior and inner parameters under the premise of ensuring the accuracy of calibration. Because the noises inevitably appear in the image, basic digital image processing technology are researched in image processing part, including gray transformation, noise elimination, histogram modification. In the research of stereo matching, the epipolar constraint of matching is analyzed, the epipolar rectification algorithm is implemented, on the basis of this algorithm, the Harris corner detection algorithm is used, and stereo matching is completed. The algorithm makes the matching time shorter and meets the requirement of the application background. The algorithms of Meanshift is researched in object tracking part, and a modified Meanshift algorithm is proposed, it enables the object to be tracked when the object is in abnormal movement condition.The experiment has shown, under the current experiment condition, the detecting system can real-time, accurate and effective tracking and positioning of specific objective.