Application Human Tracking on Electric Wheelchair using Cluster Analysis

This paper propose computer vision based human tracking which was implemented on electric wheelchair to actively follow the movement of human as guide. The main problems to be solved in this case were how to track pre-selected human and how to control the movement of the wheelchair based on the human position. In order to detect the human in a captured frame of image, we analyzed sub images of the frame using correlation. The image was divided as sub images arranged as row and column of grids. By using this method, we were able to locate pre-selected human in the image and used it as driving control rules. The computation performance of this method was better than using template matching or color-space method, which was 0.39s per frame compared to 1.5s and 0.6s respectively. We also compared the accuracy of this method with template matching and color-space method. The result also showed that the accuracy of this method was slightly better.