A Target Localization Algorithm Based On Monocular Vision

With the development of machine vision technology, the size of the object detection, moving object tracking and aircraft navigation and other aspects of space,and has great development by leaps and bounds breakthrough, and in some areas have been perfect to replace humans. For certain high repeatability conditions or requirements for precision measurement and non-contact measurement, the request for information or for quick access to the case, also, or for bad objective environment, the performance is more machine vision projecting, more appropriate.First, the article describes the classic Harris corner extraction algorithm and image edge feature extraction algorithm, based on the outline of proposed species of these two algorithms rapid extraction algorithm strong real-time, high accuracy for a relatively simple background objective environment. Simultaneous analysis of three classical image matching algorithm: SIFT, PCA-SIFT and SURF, analyze their strengths and weaknesses, as well as the conditions applicable to each.Secondly, this paper analyzes the classic camera parameter calibration method,and then using stepwise linear calibration method to calibrate the camera, and the use of cross-ratio invariance principle, namely, select an image close to the edge and collinear four feature points, acquiring an image distortion coefficient. Finally,distortion factor of the image edge distortion point is corrected to obtain the ideal image, so as to achieve simplification algorithm to improve detection accuracy.Finally, this paper based on bifocal monocular visual pose estimation algorithm.Here the use of bifocal imaging model, combined with the least squares method,calculates the distance of the object from the camera. Then he introduced the three main monocular vision targeting methods, and the analysis and study of these three methods. Finally, choose a target characteristic method using translation transform,combine quaternion theory to calculate the target attitude angle in space, and ultimately the target substance.