The Safety Protection Modeling In The Public Area Based On The Recognition Of Abnormal Behavior

Now, in many public places, such as: banks, schools, residential areas, institutions and others, The guarantee of Public safety have become increasingly demanding, some conventional techniques, for example :read identification card, manual registration, RFID with total positioning, camera surveillance, police and fully secure access platform, etc.there exits several problems: slow reaction of the security system itself, serious aging hardware, low efficiency of database research, alarm untimely. These problems affect the safety of public spaces. With the development of technology, technical staff constantly updated software and hardware equipment. In the computer field, the study of human recognition of abnormal behavior applied to the field of public safety, Security in public places has been effectively improved.Based on the human body, analysis human behavior to identify whether there is abnormal behavior, abnormal behavior will be classified by level, a higher level(if you have criminal tendencies) automatic alarm, and its backup data, medium level(abnormal behavior, but no criminal tendencies) be reminded, that low-level(normal behavior) is not processed.This article first describes human behavior recognition background and the current research status domestic and foreign areas, And then achieve the input video data, the image preprocessing(sub-frame processing: converting the video data into a continuous sequence of images) to extract human motion key-frame: a set of video include thousands of frames,finding some representative motion keyframe. Target Detection: background removal,extract human target. In this experiment, the video data is extant, doesn’t require the camera capture.Template matching algorithm in the application of a static background:the preheated human motion form image template matching library images, After extracting key frame,Filtering 、 denoising the image. Pretreatment before the match: the image isnormalized and the like. Based on match results to determine whether the abnormal behavior: the walk, waving, nodding defined as normal behavior, will significantly waving,high jump, rushing and other abnormal behavior is defined as the amplitude of the light, it will lift up sword, couples or people fighting and other definitions serious abnormal behavior. Test results are classified, high degree of matching must be considered prohibited,alarm and warn the act, the lower the degree of matching is considered suspect behavior, the management area will be tips.