抽象的

Human Tracking and Counting In Range Images by Using Mean Shift and Pylon Grid Algorithm

P. Karpagavalli, A.Vinoth Nelson, A.V. Ramprasad

Pylon grid method used for automatic tracking and counting of people on consumer path in a shopping mall, a subway and an airport in order to collect statistical data about the consumer behavior. In visionbased algorithms for the people counting and tracking has failed when the background changes gradually over time. The contour segmentation method has failed due to multistage threshold limits. If the observed area is so crowded, an occlusion problem has occurred. So we introduce the new algorithm, Pylon Grid, which is used to eliminate all these problems by using a static grid of measure points – pylons. A modified human body tracking system based on the Discrete Wavelet Transform and Mean-shift algorithm is proposed. The proposed algorithm of motion object detection and segmentation is applied to the video for a crowd of fast moving people. Tracking multiple people accurately in cluttered and crowded scenes has failed in a single view. We present multi view approach to solving this problem. This can be done with a stereo camera. Pylon grid method is the efficient and fast method for human head detection in range images captured by a stereo camera that is positioned vertically, pointing from the roof to the ground. The human head is detected using segmentation of the images. The segmentation process helps to identify the heads of the humans. The objects in the video are obtained using normal Background subtraction. The identified objects are then marked. By using the number of blobs identified the number of people in the video. An essential component of any tracking system is the detector, which determines the positions of the people in each frame of the image sequence. While the vertical camera orientation limits the size of the observed area, it provides a clear view with minimal occlusions among the people. We show how a fast median filter can be used for effective preprocessing of the range data. The person can be detected by using the method only two threshold values. The algorithm is guaranteed to find all local minima and its speed only depends on the resolution of the grid.