The Study Of Gesture-Based Human Suffring State Determination

With the development of society, there is more and more empty nester family. The problem we have to face is safety for empty nester, of which the highest risk is the elderly fall. The old man fell may lead to loss of consciousness, causing injury even death. With the rapid development of computer vision technology, Intelligent video surveillance systems, as a new and effective means of security, has been paid more and more attention by people. Therefore, to establish a method for the automatic detection of empty nester fall behavior is very meaningful.At present, fall detection existing methods have many deficiencies. Detection method based on hardware sensors, need to carry sensing equipment, and the price is expensive, can’t be widely used. At present, the fall detection method based on video is drawing more and more attention, because of the rapid development of IT technology, video technology has been rapid development and the popularization, and the price is relatively cheap video cameras. This paper has studied the decision method based on gesture. The main research contents are as follows:(1) In the pretreatment of image, this paper introduces several image preprocessing algorithms, such as binary image, median filtering and morphological processing, which is the foundation for human motion detection.(2) In terms of Detection and segmentation of moving human target, this paper summarizes the basic algorithms, analyzes the advantages and disadvantages of each kind of algorithm, and combines with the threshold method and the mixed Gauss background model method for detection and segmentation of moving human target. Moreover,this paper eliminate noise and shadow for the foreground image and further solve the "empty " problems in order to get accurately moving objects.(3) In fall detection algorithm, through the study of fall detection of existing methods, this paper presents an improved detection algorithm for the human body based on video. For some exercise and suddenly squats are false detected, the algorithm adds two methods-effective area ratio and center variation ratio-besides human aspect ratio. It improves the correct rate of fall detection, very simple and easy to analyze.(4) For improved automatic detection algorithm, in the windows environment, this paper develops a set of system for human body detection based on video, using the development platform of VC++6.0 and OpenCV. The system has good real-time performance, can correctly detect human fall, and has good robustness.