Research On Face Detection System

Compared to the traditional identity authentication, face recognition is friendly, convenient and secure. As a key link in face recognition, face detection has great significance to be studied. In recent years, Face detection technology has gained rapid progress and development, but there are still some deficiencies needed to be further improved in related research. This paper studies and analyzes face detection technology based on skin color model and the Adaboost algorithm, designs some improved strategy, and then puts forward a proposal of face detection system, finally demonstrates the good performance of the system with the experimental results.The main work and achievements of this paper are as follows:1st, taking into account of the features and advantages of skin color detection and Adaboost algorithm, this paper puts forward a proposal of face detection system with skin color segmentation initial detection and Adaboost face confirm. Besides, this paper implements a traditional face detection system with OpenCV to verificate the performance of the design in this paper.2nd, about Adaboost, on one hand, this paper proposes two improved algorithm to improve the training speed. On the other hand, this paper proposes one improved method to improve the image detection speed, and finally validates them with experiment.3rd, in the design of color image preprocessing, we propose an improved light compensation algorithm based on reference white algorithm. We can judge if the images need light compensation before processing, and make the light compensation to RGB differently.4th, about the determine of candidate face region, we select the oval color model, which can quickly segment the skin regions, after the segmentation, we should process the binary images with Mathematical morphological methods and some other methods to eliminate fake regions, while ensuring that the processing speed is high enough, we can achieve a better candidate face region segmentation.