Research Of Detecting Human Targets From Complex Background In Infrared Image

The infrared imaging has a strong penetrating power, so compared to visible lightimaging, it has many advantages. At present, the infrared imaging technology has beenwidely researched and used. The main research goal is to the human body of infraredimages and the main research direction is the human body target detection and tracking.The human body target detection is the premise and foundation of the human bodytarget tracking. And scholars now also proposed many human target detection methodsin infrared image. However, those methods have been put forward mainly aim atdetecting the pedestrian targets with simple background. And the background ofapplication is relatively simple, Such as night video monitoring and the auxiliarydriving in bad weather environment. But, in real applications, the target environmenttend to be more complex, which the target may be mutual shade or part covered byother objects. Therefore, in order to improve the practicability of the human bodydetection method in infrared image, the research on the human body infrared targetdetection method in complex background is of great significance.In this paper, the static human targets in one frame infrared image are as theresearch object to study the human body target detection technology with complexbackground. The main content of the paper include the following:①Introduce the general steps of the human body detection technology in infraredimage which are commonly used and the principle and significance of each step. Then,the characteristics of each method are analyzed. Finally, combined with thecharacteristics of these methods and the result of the commonly used segmentationmethod for infrared image with complex background segmentation, paper summarizesthe problems which are still existing in the human body detection in complexbackground.②In order to solve the problems of human body detection in the complexbackground, the paper put forwards a new human body detection method in infraredimage based on target gray search. The essence of this method is to use an algorithm tosearch the area which is close to target grey value in infrared image. First of all,according to the principle of infrared imaging and the change rule of human skintemperature, make use of statistical regression analysis method to estimate the centergray level of the human body target. Then according to the principle of clustering, this paper proposes a new target clustering segmentation method, The method takes theenvironment temperature corresponding grey value and human body skin temperaturecorresponding grey value as the clustering center, and set one third of the Euclideandistance between the two clustering center as clustering threshold. The experimentalresults show that, the new method can quickly and efficiently detect the human bodytarget in infrared image with complex background.