Recognition and Separation of Objects on a Lego Mindstorms NXT Conveyor Belt Using Log-Polar Transform and Artificial Neural Networks

This paper treats the implementation of a computer vision system to do the recognition and separation of objects that are put into a conveyor belt, built with a Lego Mind storms NXT^® kit. The image acquisition is made through a web cam upon the specific capture area. After the image have been sent to a computer, a preprocessing made by Matlab^® extracts only the main characteristics of the object, e.g., removing the background. Due to rotation and translation of the parts, a Log-Polar Transform (LPT) is used to avoid these effects. The LPT response is used as input to an Artificial Neural Network, able to classify the objects and send the information to the conveyor belt, which separates equal objects in the same place.