OBJECT RECOGNITION THROUGH KINECT USING HARRIS TRANSFORM

Abstract

With the growing research in the field of computer vision and image processing, numerous applications of Object Recognition have been developed. This paper discusses the recognition of objects through image processing. For this purpose RGB camera is being used. Initially, templates of objects (ball, box and bottle) are stored and processed for template matching. At real time original image data is taken and then divided into four frames, containing an object in each frame. These frames have resolution equal to the resolution of the templates. All the processing is done on colour images. Then, Harris Transform is applied. Using the co-variance matrix, error is calculated and compared with the threshold to match objects. Object is basically recognized by calculating the difference in the co-variance of images.