Content-Based Scene Change Detection of Video Sequence Using Hierarchical Hidden Markov Model

This paper presents a histogram and moment-based video scene change detection technique using hierarchical Hidden Markov Models(HMMs). The proposed method extracts two types of features from wavelet-transformed images. One is the histogram difference extracted from a low-frequency subband and the other is the normalized directional moment of double wavelet differences computed from high frequency subbands. The video segmentation process consists of two steps. A histogram-based HMM is first used to segment the input video sequence into three categories: shot, cut, and gradual scene changes. In the second stage, a moment-based HMM is used to further segment the gradual changes into fades, dissolves and wipes. The experimental results show that the proposed technique is more effective in partitioning video frames than the threshold-based method.