Research Of Face Object Detect And Tracking Method Based On Saliency Detection

It is well known to all that many years ago, information technology developednot so well, our video monitor depends on the professional personnel management, itneeds stared at the monitor display real time, in some important places are took shifts24hours a day. On-duty staff judge the pedestrian behavior in the video throughtheir experience,identify whether the pedestrian’s behavior is dangerous or not,do corresponding action if threaten happened. The disadvantages of this wayof monitoring are first, the large waste of human resources; Second, To judge whetherthe pedestrians behavior is dangerous or not is not scientific；Third, all-weather tookturns to guard is not reality. Modern video monitoring system with24hoursunattended real-time monitoring, judgment, early warning, and other functions intoone System, for people living, in various places such as airport, military region, banksetc; to small places such as supermarkets, warehouse, private homes and other places,bringing greatly convenient to people’s life. Most of the threats of this kind ofmonitor is a person’s behavior, so a kind of real-time, robust and accurate facedetection algorithm is the top priority.Image saliency detection is a hot research topic in recent years. The purpose ofresearching saliency detection is to extract a target which is the most important andgets the most interest in a scene. In fact, when observing a scene, human eyes tend tofocus their attention on a few goals which get most interested automatically, excludethe ones they don’t interest in. This is the human vision characteristics. Saliencydetection technology has been widely applied to many fields, such as: imagesegmentation; Target recognition; Image compression, to name a few. Saliencydetection method usually can be divided into two types. One kind is the bottom-upmethods are data-driven saliency detection method. Another kind is a top-downtask-driven method. Saliency detection via graph-based manifold ranking belongs tothe first kind. The algorithm gets good experimental results in the aspect of speed and accuracy under the background which is relatively simple. However, in thebackground of more complex cases to perform the algorithm is not able to accuratelyextract the target, the extracted results include some background information, this willhave a big impact on the advanced image processing.Face detection technology originated in the60s and70s last century, afterdecades of development, has been widely applied in biological informationidentification, security monitoring, the public security organs, and so on. Skin color isone of the important features of a face. Despite the development of face detectiontechnology for several decades, the real-time, accurate and efficient face detectionmethod has been concentrated pursuit goal by a vast number of researchers.Target tracking is always a very active field of artificial intelligence and machinevision, because it has wide application fields. Target tracking is the basis of the fieldssuch as crowd behavior recognition, people number statistics, human-computerinteraction etc,a good target tracking algorithm is very important to study these fields.As part of the pedestrian tracking field face target tracking is the cornerstone of facerecognition, authentication,intelligent monitoring and other applications to name afew. In view of the applicable scenario, whether a tracking algorithm is good or notneeds to overcome a variety of complex conditions, such as light, shade, targetrotation, etc. This article studies the saliency detection algorithm deeply, Madean improvement of this algorithm to achieve more ideal significant target extractionresults, And the improved saliency detection algorithm was applied to face detectionand has reached the experimental results are satisfactory. Finally, designed andimplemented a face target detection and tracking algorithm based on image saliencydetection. Through lots of experiments show that the algorithm overcomes the lightand shade complex conditions, can track the human face target more real-time,efficient, accurate.Provides the guarantee for face recognition, authentication and soon.