The Adaptive Detection And Elimination Algorithm Of Moving Shadow In Surveillance Video

As an important research direction of computer vision,shadow detection and elimination are mainly used in image analysis,such as target recognition,image segmentation and motion detection.Currently,if the environment changes,various algorithms that have been opposed will often be false or missed detection.To solve this problem,we proposed two robust shadow detection and elimination algorithms.(1)A shadow detection and segmentation algorithm based on image segmentation and region growth is proposed.Firstly,a gradient image compensation method is used to improve the existing Mixed-Gauss model,extract moving target image,eliminate image hole,noise interference and other problems.Then,the moving target and shadow region are segmented by graph-based image segmentation.Finally,the regional growth algorithm is adopted to merge the segmentation regions,and the moving target and the shadow region are segmented accurately.(2)A shadow detection and elimination algorithm based on gradient and YUV color space is proposed.Firstly,the moving target is extracted by using the PBAS motion target detection algorithm,and the Ghost area is suppressed by increasing the frame difference method.Then,the gradient map of the moving target is compared with the corresponding background gradient map,and the foreground target candidate region of the shadow region is obtained.Further,the shadow detection algorithm based on YUV color space is adopted to detect and eliminate the shadow region of the moving target and obtain another part of the candidate area.Finally,through the characteristics of the shadow,the two parts of the candidate region are combined together to obtain the moving target of eliminating the shadow region.Compared with the existing methods,the experimental results show that the two kinds of shadow detection and elimination algorithms proposed in this paper have better robustness to different environments.