An important component of video communication system is video compression, which is the art of reducing the size of video file without the loss of visual quality. Apart from lossy and lossless compression, to increase the performance of compression algorithm, a ROI (Region of Interest) based compression algorithm is used. In ROI-based compression algorithms, a lossless algorithm is used to compress ROI region and lossy compression is used for non-ROI regions. In these types of algorithms, an important step is the accurate separation of ROI and non-ROI regions. In this research wok a region separation algorithm that extracts the ROI region effectively from its background is proposed. The ROI considered in this work is the human face. The proposed algorithm uses a dynamic hybrid color space algorithm along with Ant Colony Optimization and Kernel Principal Component Analysis algorithms to detect face skin regions, followed by the use of eye-mouth map features to detect face region in the video. A Kalman filter is used to track the detected ROI in the video. Experimental results shows that the proposed algorithm is efficient in identifying the face regions in terms of accuracy.
CITATION STYLE
Ramadoss, M., & Mahendran, S. K. (2019). Design of enhanced region separation algorithm for effective video compression. International Journal of Recent Technology and Engineering, 8(1), 2374–2379.
Mendeley helps you to discover research relevant for your work.