Research On Human Face Detection And Image Enhancement Of Surveillance System

In the use of video surveillance system, the images have the weakness of image fuzzy seriously. So, the human face detection and image enhancement research in the surveillance system get great importance over the recent years. Paper do the research on human face detection and image enhancement. The main work as follows:In the pre-processing stage, using of image subtraction operation to make the target person image being segmented, that reduced the amout of calculation in the next step.The SSR filter algorithm is based on characters with a combination of template matching method. The principle is that using the relationship between light and shade of the face to look for face candidates. Improved two points, one, using the length and width ratio 2.22:1 instead of origin SSR filter ratio 2:1, this is a statistical result from many standard face images; another,improved the average"Between-the-Eyes"image template matching algorithm, and according to the matched degree value to determine whether the candidate is a true face region. In the experiment, the standard front side face detection rate makes 95%, the face candidates verifying rate makes 92%; the surveillance system image gets face detection rate of 90%, and the verification rate reaches 80%.Some methods as space domain transformation, space domain filtering, frequency domain enhancement were adopted to the detected face candidate images to make enhancement processing. After that, the images have more clearer visual effect.Experimental results showed that the improved algorithm has better detection rate than the original. It has laid a foundation for the further study.