A novel image edge detection method for workpiece based on improved Extreme learning machine and information measure

Abstract

The edge detection and extraction of workpiece images is one of the important links in industrial automation production process. Image edge detection accuracy of the workpiece of traditional edge detection operator is limited and high sensitivity to noise, aiming at these problems in the detection link, a novel image edge detection method of workpiece based on the improved extreme learning machine is put forward. The method constructs a description vector of the characteristics of the edge points information measure, use the training sample which composed of feature vectors to classify the improved extreme learning machine to realize the edge detection. Experimental results show that the new method proposed in this paper is better than traditional Canny operator, Sobel operator and the method of ELM image edge detection, and the extracted image edges are more complete and clear, with stronger noise suppression ability.