People Counting Based On Video Image Processing

Along with the developing of this society, people counting has been widely applied in the station, the path and the supermarket. The paper mainly studied on the video image processing people counting based on a lot of correlative data. On the research of image processing, we mainly carry on the conception, principle and characteristic of the noise removal ,image sharpen ,image smoothness and segmentation according to the purpose of the paper. Mathematics morphology filter is also applied in the article and meanwhile, the basic principle and fundamental operation are also described.People counting in gallery has been researched in this paper. In the processing of counting people, the hair region is extracted by using the gray value and the proper threshold of pixel value is set to estimate the real head region. An adaptive threshold is used here. For the people counting in gallery case, the maximal gray value is first compared with the threshold of pixel value in the subtractive image ,in order to estimate the interference whether it is noise or not. If it is not absolutely interference of noise, the adaptive threshold of pixel value will be chosen for the subtractive image and the position of centroid will be recorded. The people counting on the condition of just appearance, just disappearance and still in the image will be judged by the initial direction symbol bits, the symbol bits of changed direction and the current position.The contribution of this paper is introducing a new method instead of the tracking method to reduce the requirement of the sample frequency. Adaptive threshold is used in the paper. The choice of adaptive threshold is useful for overcoming the influence of the light too bright or too dark and effectively adapt to the case of sunshine or the night. It is shown in experiments that the person number can be calculated accurately by the method.