DefinePK

Abstract

Face recognition has been one of the most important areas of research in the field of computer vision and video pattern recognition due to its broad range of profitable and a law enforcement request. The highly realistic demand on face recognition remains a subject of extensive research. The face recognition has shown success for high quality images under controlled circumstances, but video based face recognition is hard to attain analogous level of performance. The work region of key frame extraction is so extensive and effective technology. Many techniques for video key frame extraction have been report in so far. The existing 3-D face models provides enough viewpoint variation to carry out stereo motion but was not effective with single ultra high resolution camera for key extraction on both wide and narrow angle tasks. To overcome the above mentioned issues, Key frame Extraction using Wavelet Information (KEWI) scheme is developed to use in video summarization. In KEWI extracting key frames, two consecutive frames namely the Discrete Wavelet Transform changed and then the differences of the detail components (i.e. eyes, scars, moles in the face) are estimated. In KEWI, if the diverse value of successive pair is superior when compared to the threshold value, the last frame of the pair is considered as a key frame. Experimental results show that the KEWI scheme easily detects the images using the ground-truth dataset of 1000 videos with expected objective group labels. The experimental performance of KEWI scheme is evaluated in terms of discriminative level, energy consumption, cumulative accuracy, verification rate and key extraction of face region efficiency.