Neural networks for bar code positioning in automated material handling

Abstract

This paper presents an effective method to utilize the specific graphic design of bar codes for positioning objects on conveyor belts without work carriers. A simplified template matching method is utilized to detect the four corners of a bar code. After the four corners are located, an artificial neural network is utilized to acquire the translation, orientation, and vertical depth information of a workpiece for the bar code scanner and robot workstations. This proposed system is successfully implemented in a low cost computer vision system for automated material handling.