Research And Implementation Of Human Face Detection Based On Davinci Platform

The human face detection technology was come originally from human face recognition. Since the sixties of last century people began research human face recognition. Nowadays the human face technology has been developed more maturely. Human face detection is a crucial step in Auto Human Face Recognition. At the first times, the researchers were mostly study on strong restriction human face images, there were no background in the images, they assumed the human face is easy to retrieve, so the human face detection problem was not get enough attention.Nowadays, along with the development of the applications in E-business and security supervision, human face recognition has become one of the most powerful ways to get biological identification. In these applications, the system needs to get human face area in a complex background image first. Then, the human face detection problem became an independent subject to be research on and more researches are working on it.The definition of human face detection is to confirm the human face in an image, and use some kind of searching strategy to find the human faces'locations in the image, it also need to get the face size and gesture. The face detection technology is the first step of a serial modern artificial technology.Along with the research of face detection, a lot of algorithms were come out. And the AdaBoost, first proposed by Freund, is the most powerful one. It is consist of serial weak classifiers. These weak classifiers can automatic build a strong classifier. Viola and his colleague improved AdaBoost to fit the face detection; the most important method they proposed were Haar feature. With their improved algorithm, they build a waterfall-like front human face detector, and it had a high accuracy and high speed effievence. This method has been proved one of the most powerful one.DaVinci technology is a great performance Embedded platform. It is a suitable platform for digital multiplies media. DaVinci technology was proposed by TI, and it consists of CPU, software, IDE, support and server. This technology has a great feature, which combined the high speed computation of DSP core and the powerful peripheral management, it make the DaVinci can fit every video/audio application. Apply the human face detection technology on embedded platform is a tendency on life applications. Now we can put it on the powerful DaVinci platform.This paper study and make a human face detection system on DaVinci platform. To make the system meet the real-time requirement, we optimized the algorithm and also made some programming improvement. Programming on DaVinci is different from normal CPU; the software will not only run on DSP and also on ARM, so we have to build both two sides programs. DaVinci technology has a serial media programming standards, such as xDAIS and xDM. So when build programming we also need to fellow these rules. And make the two side core collaborate well.