Research Of Safety And Energy Saving Detection Technology Of Automatic Door Based On Omni-directional Vision Sensor

With the rapid development of the city modernization, the automatic door is widely used in hotels, airports, shopping centers, banks, office buildings and other large public places because of its advantages of good appearance, convenience and energy saving. In recent years, when applied to human’s daily life, automatic door was found to have efficiency and safety problems. The reason is that the traditional sensors of automatic door have three main flaws listed as follows: 1) the information detected is single, so they could not meet the requirements of safety, energy saving and comfort at the same time; 2) there are dead angle in the detection range, so it’s difficult to detect people or object who was under a certain height; 3) they are lack of intelligence, we could not analyze and understand pedestrians’ behavior with the traditional sensors. For this reason, a safety and energy saving detection system of automatic door based on omni-directional computer vision and intelligent video analysis technology is proposed.First of all, the classification and characteristics of the existing sensors of automatic door are summarized. By analysis of the research about applying intelligent video analysis technology to automatic door system done by domestic and overseas scholars, the problems to be solved are put forward. On the basis of these problems, a safety and energy saving detection system which obtained information from the omni-directional vision sensor(ODVS) is designed. Firstly, the video image data collected is analyzed and processed. Then, the motion status and behavior of pedestrians around the automatic door are recognized. At last, intelligent control signal will be sent to automatic door system to realize the performance of safety and energy saving. The research is mainly focused on the following aspects:(1) The pedestrians around automatic door are detected and tracked by the MHoEI(Motion History or Energy Images) algorithm through the omni-directional video image captured by ODVS without dead angle. In the experiment, the performance of foreground detecting and shadow removing of MHoEI algorithm is compared to which of the widely used Mixture of Gaussians background subtraction method. Tests proofed that MHoEI algorithm can not only do well in foreground detecting and shadow removing but also save more computing time and storage space for systems.(2) Through the motion information obtained by MHoEI algorithm, the motion status, safety status and abnormal behavior of pedestrians could be analyzed to identify their activities. Then, intelligent controlling signal will be sent to the system of automatic door.(3) Realize two kinds of people counting methods, one is based on the area of region and the other is based on the features of head. And a comparative performance analysis on speed, memory requirements and accuracy was presented.In this thesis, a safety and energy saving system which is friendly to human beings is designed and realized using Java language. The main modules and related methods in such a system are introduced in detail. The last part of the thesis is used to summarize the research. Experiments show that the proposed system not only has characteristics of convenient installation, simple maintenance, intelligent operation etc, and also can detect the pedestrians around automatic door effectively, analyze the normal and abnormal behavior of them correctly, monitor the operation of automatic door in real time, achieve the function of people counting accurately.