Research And Realization Based On Digital Image Processing Of Face Detection Algorithm

Computer face recognition is very active field of research in recent years. Its applications is wide, such as identity authentication of security system, the target identification and tracking of video monitor , and facial expressions analysis, age analysis, lip reading, etc. Face detection is based, it is key in automatic face recognition system, effective and accurate face detection algorithm has a very important role in improving face recognition rate.This article mainly discussed face detection problem in color facial image, want to undertake face detection, main need to carry out facial image denoising and edge detection and face image segmentation, processing of illumination ,ultimately you can tell a face, aim at facial image denoising ,proposed one new optimized wavelet threshold value contraction law algorithm based on the fast particle swarm, particle swarm optimized algorithm was used to extract the threshold value optimal solution, then particle swarm optimal solution is used as the wavelet decomposition each criterion threshold value, carried on the image denoising by this most superior threshold value solution, the experiment proved that the method, not only PSNR is obvious enhancement, moreover the picture quality and vision are also improved, and is bigger along with the noise variance, the PSNR and image quality is better.because of the traditional edge detection inadequacy , use quaternion's vector rotating theory, improved Particle Swarm Optimization algorithm is introduced to quaternion edge detection, we present a new color image edge detection method. Experimental results show that the proposed method of color image edge detection has good effect, it extract the image detail that many traditional methods can not do so, and the algorithm is stable, easy to convergence, edge detection a little faster.Aiming at the common image segmentation shortcomings of large amount of calculation, the disadvantages of segmentation effect, this paper puts forward a two-dimensional Otsu method. first image is divided into n×n window, in each window two-dimensional colony algorithm is used to search the optimal threshold vector, each bee represents a possible two-dimensional threshold vector, often through flying for each bee get the optimal threshold. Simulation results show that the proposed method can not only be good segmentation results, and the calculation is greatly reduced, achieve a fast segmentation purposes, facilitate real-time application of two-dimensional Otsu method, and thus the method is feasible and effective. finally light processing method of face image was introduced.