The Comparison of Object Recognition and Identification by Using Image Processing Base on the Neural Network, the Hough transform and the Harris Corner Detection

Abstract

This paper presents an object recognition and identification system using the back propagation neural networks. The performance of Hough Transform and the Harris Corner detection are compared with the following procedures and methods; the webcam is used to capture the object and create an input image, change the color image from RGB to gray scale, resize, learn and recognize the objects by neural network, and separate the objects by the robot arm. Three different types of objects in this study are triangle, rectangle and rigid circle. The object recognition and identification from the neural network, the Hough transform, and the Harris corner detection are compared. The results showed that the neural network gives more accuracy than the Hough transform and the Harris corner detection.

Recognition of Geometric Shapes in Images

“…It recognizes that the line is evaluated at the point on the plane x, y. Likewise, CONVENTIONAL HOUGH TRANSFORM (CHT) can find a curve and a circle as given in equation (3).…”