Abstract
Face extraction in images and video is very vital in security and internet communication. Face detection is a process of separating human faces from non faces objects in an image or a video frame. It is the basis of face recognition and other biometric applications like video surveillance and tracking suspicious activities. The goal of this project is to first simulate and then implement the face detection algorithm on Tl's DSP based Video Development platform DM6437 DVDP. Computer based simulation uses RGB and HSI color space for the determination of skin in images. For the hardware part of the project, a frame is separated from video and processed to detect any human appearance in that frame. The steps I used are human skin separation from other parts of image, connectivity analysis method for the region detection and face feature localization. The technique is based on YCbCr color space processing. Individual frames from video are searched for human appearance and the results are used for face tracking. Finally, the presented results of the project demonstrate the successful implementation of the face detection and face tracking using the DM6437 platform.