An FPGA Based Facial Component Extractor using Wavelet-Based Salient Points

Facial component extraction is an active area of research and numerous methods are available for it. Some of these methods are slightly inaccurate because of vast dynamic characteristics of human face and its shape variations. However, these methods are widely used in real-time applications due to their low computational cost. In this paper we propose a new method that uses wavelet decomposition to extract salient points of the given image. Statistical analysis of the obtained salient points can exactly locate facial components. The proposed method tries to achieve computational efficiency and accuracy simultaneously. Simulation results present a detection rate of 95.6% for facial components. A complete FPGA implementation of the system is also proposed. It is shown that the logic resource usage for entire system is so low that the entire system can fit into a 300K gates Sparatan-IIE FPGA. The designed hardware is capable of extracting facial components for up to 5 frames per second