Research On Moving Human Positioning In Complex Background And Fixed Camera

With the development of economy and the improvement of people’s living standard, moreand more people begin to pay close attention to the security industry. In the security industry,video surveillance technology has great potential of development and is one of new scientificresearch field. In the applications, video monitoring technology range is wide. Moving humanpositioning is one of core technology in the intelligent video surveillance system.Because the fluidity of the natural environment, for example, change of light intensity,moving objects of similar skin color, moving objects without skin color and moving objectsof similar circular, blocked the development of moving human positioning. With the researchdevelopment of moving human positioning, the single moving positioning algorithm hasmany deficiencies, such as detection by leaking and mistaking.To solve the disturbed problem from complex background, a moving human positioningmethod which fuses the human skin color, the environmental illumination, the head contourand the human prior knowledge is proposed in the complex background.First of all, this paper introduces some basic knowledge of the image preprocessing,including images of gray, image filtering, image binarization, morphological filtering, edgedetection and so on.The improved frame differential method extracted the moving area from the videosequence image. The improved frame differential method can avoid the unpredictablecondition from traditional frame differential method at t+1.Again, the human skin color regions with the improved frame differential are extractedwith the self-adaptive environment illumination method. The image brightness mean bycomputing is the characterized parameters for blending into skin model. The method caneliminate the influence to position human skin color from the environmental illumination andsome similar human skin color moving objects combined with prior knowledge.Finally, according to the head shape is the similar circle, the moving human is position. Inthe moving regions without skin color regions, adopt the improved Hough transform toposition similar circle contour regions, and to determine the regions are whether for movinghuman regions. This paper use prior knowledge to reduce spatial dimensions of theparameters in Hough transform. In PC, using MATLAB and C programming detect the moving human in images ofshooting. The algorithm compared with other algorithm in the experimental results, and theexperimental results show that the algorithm has the better effect than the algorithm using thesingle feature to location moving human in the complex background.