Video shot cut detection using least square approximation method

Abstract

Video shot boundary detection is the basic step in the area of content based video analysis and retrieval. Various automatic shot boundary detection techniques have been proposed and their performances are reliable, especially for video cut detection. However most of the proposed methods are sensitive to camera, object motion and lighting changes. In this paper, we have focused mainly on the improvement of the existing traditional shot detection methods. In our technique, approximation of the discontinuity values obtained from the various existing methods has been done using the Least Square polynomial approximation method to diminish the sensitivity to motion. An automatic threshold calculation algorithm is also used in our method. Experimental results demonstrate that our method can detect maximum shot breaks and reduces the occurrence of false hit that are difficult with the previous approaches when motion occurs in the video. © Springer-Verlag 2011.