원문정보
초록
영어
Shot boundary detection is used to summarize the video in the video content management system so that particular content or required scene can be extracted from a given video and also observers or users can easily find summary of the given video in short time instead of time spending in seeing full length video. Also shot boundaries can be used for video indexing, browsing, and retrieval. So to extract such key frames (shot boundaries) from a given video needs a robust technique with high accuracy and low error rate. In this research work, we proposed a robust technique to find shot boundary using feature vectors, which are obtained from features calculated for each sub-band in the contour let transform domain. The feature vector has the energy, standard deviation, and histogram similarity as the features of each sub-band. The experimental results proved that this novel and robust method had more accuracy rate and low error rate.
목차
1. Introduction
2. Overview of the Proposed Work
3. Contourlet Transform
4. Feature Vector Computation and Classification
4.1. Feature Vector Computation
4.2. Classification by Euclidean Distance
5. Results
6. Conclusion
References