Research And Implementation Of Camera Track’s Analysis Self-adaptive Algorithm Based On Monocular Vision

At present, computer vision is widely used in visual inspection, visual navigation and automatic assembly field. In visual system, camera is the tools of shooting, and the shot objects are processed. However, as a passive sensor, if the environment becomes complex and severe, it is not suitable for human to manipulate camera, the working range is great limited. In order to expand the camera’s using range and realize the camera’s autonomization and mobility, the image resuorces obtained by camera can be fully used to make the camera autonomously return and guide independently. It is the urgent demand of realizing the camera’s automation to stuy the ability of autonomous return.The camera moves in the3D space without GPS navigation or ground guidance information. To realize independent return, the motion track should be known. That’s the main work in this paper. A series of pictures without sequence are clllecioned by camera. According to the global motion which is caused by the camera’s motion, the relative motion information of camera should be accessed to recover the camera motion route.First, the relationship of the camera’s motion and the points’ motion in the image is reserached. First, the study of how to get motion parameters information from image is researched, choose accutatly located points as matching primitives. To select SIFT which the number of feature points’s num is large and stable as the image matching method and use the point selection algorithm based on reverse matching and edge feature to remove the mismatching points. Thereafter, obtained to the feature points set, used the robust RANSAC model parameter estimation method to solve the global motion parameters and selected the parameter model which is based on the affine transformation model to describe the movement. The global motion estimation process has been improved. Subsequently, since in the process of match using the whole image, the points’ number is so large that need large amount of computation, the adaptive sampling image matching strategy is proposed. In this strategy, the number of block and sample proportion should be controlled in order to get appropriate number of feature points which can simplify the time of feature point extraction and matching when the accuracy of parameter is ensured. The test results show that the adaptive sampling scheme which is based on four-block has been ensured as the best scheme, it can improve the efficiency of image matching when the accuracy is ensured.