A Face Detection System Design And Implementation Under The Complex Environment Based On Adaboost

Face detection is to detect human faces and provide the exact coordinate of each face in still images or video sequences, regardless of different location, orientation, size, pose, lighting condition. As a key technology in human face processing, face detection is of great importance in the field of security protection, ROI-based coding, content-based image retrieval, automatic video surveillance, human computer interface, etc. Now, face detection is one of the most active research fields of pattern recognition and computer vision.With the great efforts taken by researchers all over the world, face detection can now achieve a usable detection rate and speed. However, human face is a nature structure with highly complicated variations in detail, which bring great challenges to the performance of detection algorithm. These kinds of variations lie in pose, facial expression, partial occlusion, lighting condition, rotation, etc. Besides, human faces are always compounded with a complex background. Due to all of these difficulties, there is no such an algorithm that can handle all these variations without any kind of limitation at present.The cascade classifier is trained based on Adaboost algorithm, constructs a face detection system .The following methods are adopted to improve robustness and speed of this system: the pictures is preprocessed by image process technology. zooming, light compensation etc.; the rapid computing method of Integral Image are used; increasing the number of face pictures and add as many different skin color, different background scene, different illumination light pictures as possible to improve the accuracy of face detection; the method of zooming out the windows to detect ease the computing task compared with the conventional method of pyramid. Finally, it presents some new features and proves that the new are very efficient in face detection. Accuracy became greatly, at the same time, reduce the detection time .this system can meet the requirement of real-time detection.