The Realization Of Head Detection System In Practical Environment

In some specific scene, such as the building corridor, the school library, bus or other public places, the monitoring system is used more and more widely. But most of the monitoring systems are still in semi-artificial style, and these systems cost a large count of manpower and material resources, which hindered the popularity of video surveillance in public places. Therefore, we need a more intelligent visual surveillance system.Computer intelligent video surveillance system (CIVSS) is one of the new arising high-tech application fields, including computer science, computer vision, image processing, pattern recognizing and artificial intelligence. The system detect, recognize, and track moving objects in a special environment in real-time. Furthermore, it can also analyze and judge the behavior of objects. The aims of CIVSS are to understand the meanings of video stream and explain the scenes comprehensibly. And then we can take actions or make some decision based on these results.As to computer intelligent video surveillance, there are still lots of problems in theory research and in applications. More and more researchers devote themselves in the area and there are many important achievements for us. While head is not so easily blocked as the body and it is typical. The main research of the paper is about the technologies about the human head object detection and tracking.The main work includes the following aspects:Firstly, head detection technologies are investigated. Several head detection algorithms generally used are introduced, and the performance is analyzed. Generally speaking, researchers apply two methods to detect human heads. One is using image processing methods, for instance, some articles try to detect heads using head shape features or skin color features. However it may seriously affected by external environmental such as light condition. What’s more, it usually takes too much time to apply it in real-time cases. Another one takes advantage of machine learning method, which is widely used in recent years. You have to collect large amounts of data including positive samples and negative samples. And then choose a training algorithm such as SVM and a kind of feature representation to train a head detector. In many cases the latter one detects more quickly.Our detection adopt the machine learning method. First a large scale of positive and negative samples are collected. Then we calculate the integral channel features of the samples and train it by Soft Cascade which is based on AdaBoost theory to achieve classifier.In practice, the surveillance camera monitors the surveillance scene statically during most of the time. Thus, the background of the region is relatively static. Based on this characteristic, this paper does the pretreatment on the surveillance video at the beginning, by adopting ViBe extract the moving foreground. There are two advantages of the pretreatment:Firstly, it could narrow the detection area, speed up the detection process and reduce the computational complexity. Secondly, the impact of background on the test results can be eliminated effectively, while the false detection rate is reduced.Then the head tracking algorithm is investigated. The paper introduces the STC(Spatial Temporal Context)algorithm and its application in object tracking in detail.Our experiments shows that the method this essay suggests can detect heads in any angels accurately in real environment at a acceptable speed.