Face Detection Method Based on SSR Algorithm

摘要

In order to achieve high speed and high reliability of face detection system, a feature-based template-matching algorithm to detect a key part between the two eyes used. The method is achieved through divided a filter into six rectangular. The face detection process adopts the "integral image" theory. Just calculating the information of around the eyes, eyebrows, and nose area, aspect ratio of the value of the SSR filter is set to 2.22:1 through the experimental which improved the SSR algorithm. Using the advantages of gray-level information, it has better reliability on the changes in light conditions. The experiment result shows that the accuracy rate of the standard frontal face detection is 95%; to monitoring images, the accuracy rate of the "eyes - human face" window location detection is 90%.