Parallel Implementation of Face Detection Algorithm on GPU

The development of high resolution digital cameras for recording of still images and video streams has had a momentous influence on how communication and entertainment have developed during the recent years. Processing of human faces discovers many applications in various domains like law enforcement, security surveillance etc. A standard face processing system consists of face detection, face recognition, face tracking and face rendering. Face detection is one of the most premeditated areas of computer vision in image processing not only because of its thought-provoking nature but because of its broad continuum of applications which requires face detection as the first step. The formative Viola-Jones face detection algorithm is discussed in detail and what is the future prospectus of the face detection techniques. The focus is on the benefits and the drawbacks of the Viola-Jones and why it is the most prominent face detection algorithm and how it can be further improvised in order to meet the needs of today's time. To overcome the cons serial implementation of the Viola-Jones algorithm, the algorithm will be implemented on a parallel platform. The primary objective is to increase the computational speed of the algorithm which may be achieved through parallel implementation using CUDA and OpenCV on GPU (Graphics Processing Unit) and then give comparative analysis of the better computational results between the serial and parallel implementations.