Object Segmentation And Tracking Algorithm Based Geometric Active Contour Model

The motion analysis to a moving object in sequential images is an important part in computer vision technology. The motion analysis consists of moving object capturing,moving object segmentation,object tracking and object understanding. The object capturing, segmentation and tracking provide the foundation for motion analysis. And a well-segmented and tracked result will lead into a correct object understanding. In recent years , geometric active contour model based on the level sets developes greatly in the image processing domain. In order to allow computer vision system automatically segmenting and tracking the object continuously and at the meantime automatically gathering the object contour information, the research has been concentrated on the techniques of moving object segmentation and tracking based on active contour model in sequential images.In this thesis , curve evolution theory and level sets method are first introduced , and the theory of geometric active contour model based on curve evolution theory and level sets method is discussed deeply, analysis its advantage and shortcoming, a new improved level sets method is introduced the complex of calculating is reduced and the speed of calculating is improved greatly by this method. Then the application of geometric active contour model is studyed in the moving object segmentation and tracking .In the process of segmenting the object, in order to conquer the shortcoming of previous image segmentation method, for example, the edge of moving object segmented is not successive and low accuracy, a new image segmentation method based on moving information of object and geometric active contour model is proposed. First, the image differential technique is used, namely calculating cumulate differential image of three sequence image. After the rigion of interest (ROI) has been es- tablished, the external rectangle of the rigion is regarded as the initial contour for the geometric active contour model. Then, using the algorithm brought forward in this paper, it ensures that the contour convergence on the edge of object exactly, obtaining the exact contour of object.In the process of tracking the object, because the position of the moving object is different in different frame, the main problem of geometric active contour model is estimating the initial contour. In the paper , we analyse the character of tracking algorithm based on Kalman filter, In the process of tracking, we combine Kalman filter with geometric active contour model, obtaining initial contour for geometric active contour model using Kalman filter.The effectiveness of the proposed algorithm of segmenting and tracking object has been carefully verified in the tests. It has been shown in the tests, through the process of object segmenting and object tracking , that the proposed approach can effectively carry out accurate object segmentation and tracking.