Method for adaptively detecting and eliminating shadow in video segmentation

Abstract

The invention discloses a method for adaptively detecting and eliminating a shadow in video segmentation. The method comprises the following steps of: firstly, roughly detecting a moving changing region by using accumulated frame differences and constructing a motion template by using a static index; secondly, performing statistics on brightness information to establish a background model, performing updating by combining the motion template, and detecting a foreground and an edge of the foreground by using a background difference and a Sobel operator; thirdly, performing horizontal projection and vertical projection on a detected edge image respectively, performing statistics on the distribution condition of edge images and roughly determining a shadow position and a search direction by combining shadow characteristics with a spatial position; and finally, precisely positioning a shadow point along the search direction in a suspected shadow region by using a hydrometer method so as to precisely eliminate the shadow. Due to the combination of the characteristics of the edge and the spatial position of the shadow, the method for adaptively detecting and eliminating the shadow disclosed by the invention has the advantages of adaptively positioning the shadow region and precisely eliminating the shadow point under the condition of invariance and availability of chrominance, along with small calculation amount and high robustness.

Description

Technical field: [0001] The invention relates to a processing method in video segmentation, in particular to a method for detecting and eliminating shadows by adaptively fusing shadow edge characteristics and spatial position characteristics. Background technique: [0002] The second-generation video coding standard represented by MPEG-4, whose object-based coding and interactive functions make semantic video object (VOP) segmentation a challenging problem, has gradually become a research hotspot in the field of video processing. As a key link in video processing and video analysis, video object segmentation technology is not only widely used in the fields of pattern recognition and computer vision, but also more and more important in emerging fields such as video retrieval, video coding, and multimedia interaction. [0003] In many applications such as intelligent video surveillance, traffic monitoring or video conferencing, the shadows produced by moving objects bring grea...

AI Technical Summary

Problems solved by technology

On the basis of color segmentation, Pan Xiang performs shadow detection by combining blue band information and HIS spatial information. This method cannot separate black objects and shadows, and the accuracy and robustness are poor (see Pan Xiang. Based on color information and edge features Motion shadow detection [J]. Journal of Zhejiang University (Engineering Science Edition), 2004, 38(4): 389-391, 407.)

Images