关键词: 电梯群控, 人数统计, Hough变换, 人头轮廓

Abstract: The elevator group control dispatch algorithm, which considers the numbers of people in elevator car, can find better solution in the limited time and is very important to the elevator dispatching. According to the instable accuracy and lower speed produced by the correct people counting algorithm based on vision, this paper proposes a counting algorithm for the numbers of people in elevator car based on improved Hough transform which is used to detect head contour. This algorithm converts three dimensional space based on the circle detection by standard Hough transform to one dimensional space, and adopts the geometrical character to traverse the image data, then accumulates value of radius by the one dimensional Hough transform, finally determines the radius and center of the head contour by comparing the values of radius with the threshold. The results of experiment show that the algorithm greatly reduces runtime and improves accuracy, has good anti-noise performance in contrast with the traditional method of head contour detection by Hough transform.

Key words: elevator group control, counting people, Hough transform, head contour