Research On Human Perception And Localization Based On Scanning Infrared Information Processing Technology

In recent years, with the development of intelligent household, human perceptionand positioning technology is becoming a research focus. Infrared radiation emittedby human body itself can be detected effectively by infrared sensor in traditionaltechniques, and human target location information can be obtained by analyzing thefeature extracted from the infrared signals. However, other non-human heat sourceswould radiate electromagnetic wave and cause interference on the human bodypositioning. Therefore, effective solutions must be adopted to exclude non-humanheat sources and increase human body positioning accuracy.In this thesis, a synchronous scanning discrimination system with two infraredsensors was designed. Firstly, Omron D6T sensor was used to detect infrared radiationof the human body so that the position information of heat source was obtained bydual-threshold and morphology processing on the eight-channel spatial temperaturedata in a loop, then RE200B pyroelectric infrared sensor which surface mounted by8120Fresnel lens was used to acquire pyroelectric data of heat sources, such ashuman body, electric equipment, light and so on, to achieve classification andrecognition of the target human body. Two feature extraction methods of pyroelectricsignal from different heat sources were studied. Fast Fourier Transform was chose asthe first method of feature extraction and the spectrum information was obtained.Wavelet Packet Analysis was employed as the second method of feature extractionand got both wavelet packet energy and wavelet packet entropy from thedecomposition in three layers. In recognition process, fisher linear discriminantanalysis and support vector machine (SVM) were employed separately and therecognition results from these two methods have been compared. The experimentalresults showed that compared with SVM, fisher linear discriminant method had betterrecognition performance for the target human body, which the lowest recognition ratewas75%, and the highest could amount to100%.A human perception and localization system has been designed successfully inthis paper, which combines two kinds of infrared information to achieve different heatsources classification according to high positioning accuracy of the heat sources. Thiswill provide a low-cost intelligent home human positioning control solution systemfor central air conditioning and other intelligent household equipment. Currently, research in the field of human body positioning based on multi-channel infraredinformation is still in its preliminary stage. The results of this thesis will promote thedevelopment of this technology to a certain degree.