Research On Object Localization In Color Video Sequences Based On Active Shape Model

In the research field of image processing and computer vision, object localization in color video sequences is one of the most active research topics in computer vision at present. Wide application has been conducted in many fields.Active shape model (ASM) proves good performance on object localization in gray level images. The model shows its better robustness compared to other localization methods because of the use of the priori knowledge. But the localization performance depends on initial position excessively. A localization method in color video sequences based on ASM is presented for moving object localization in this paper. According to human vision character RGB color space is converted to HSV color space. Instead of the gray information, the hue information is used as the model parameter. After construction of the prior contour models based on principal component analysis (PCA) technique, structures of objects can be located by adjusting the model parameters in a learning way that it minimizes the distances of corresponding contour points between model and object. A good overall match is obtained in a few iterations. Because the difference between moving object in successive frames is unobvious in color video sequences, the last position of locating in the former frame just turns into the initial position of the latter frame which solves the problem of determination of initial position. Finally through comparison of test results based on the moving fish in color video sequences, the method presented in this paper combining with the hue information shows effective results for localization in color video sequences. And the determination method of initial position presented above improves object localization efficiency remarkably, at the same time a good robustness in partially occluded object localization in color video sequences has been acquired.