A Dynamic Human Face Detection Method Based On Video Array

As popularization, improvement of performance and the studying of pattern process and pattern-recognition field that the computer employs being ripe progressively. The application that the human faces detection is more and more extensive. Some important financial institutions such as stealing and destruction that the national treasury of the bank and some departments that preserve classified paper materials suffer the criminal again and again in the society now, therefore cause serious losses. How to satisfied safer problem needing under these environmental conditions badly being solved of guarantee. Among them the dynamic face detection becomes a key ring while carrying on and controlling the department safely. It can monitor the abnormal personnel who entering and note down through dynamic person's face.We have studied this in a large amount and have proposed a kind of detection algorithm based on dynamic faces of people of video array. This algorithm includes three parts, the moving detection, skin color detection, front faces of people detection. This algorithm measures the body in motion fast through the method based on that color curve proportion difference .Human face that be cut apart with based on the structure of face, and then distinguish dynamic human face on this basis.It is the difference shadow law to measure the body in motion in video monitoring system, but the difference shadow law while measuring objects can cause noise. Because illumination brings influence in detecting the issue of the result. We accord to the value of color in the picture and value of grey level and the variation tendency is identical, the characteristic that the proportionate relationship is not changed much by the illumination. The projection difference part of two pictures orients the body in motion. Using this kind of method which carries on fast and detects accurately to the body in motion .It solves the noise problem that the illumination change brings at the same time.We sets up a skin color model to detect skin color. The threshold value comprehensive YIQ and and YUV color space .It chooses a kind of strict threshold value range, then can make people skin color screening finer.The course of human face distinguishes is base on comprehensive many kinds of methods. This method includes the mosaic of the picture, the horizontal line detection, and uses with regular structure and characteristic of human face. It can distinguish the faces of people and not human face.A dynamic human's face detection method based on video is effective , and color curve proportion difference algorithm can measuring the body in motion rapidly, the skin color models set up follow from experimental data which is basically met under different illumination conditions . The search method of human face characteristic is accurate and effective too.