A computer controlled visual system for object classification

Abstract

In this paper the design, implementation and application of a computer controlled camera system for classification and assortment of various objects is introduced. The system, which consists of a RF-Color Camera, a PC and pneumatically actuated robot manipulator, is configured to operate in real time and on line applications. On the first stage of the work, the captured image (in RGB scale) viewed by the camera, is converted and stored by using a frame grabber card and special software written in C++ into a digital image data in the computer memory for a further segmentation process to recognize the object. In the second stage, the objects are recognized in size or property via camera, which is mounted to a horizontally moved belt driven with a servo stepper motor. The recognized objects are then assorted by a built up, computer controlled electro-pneumatic robot manipulator with four degrees of freedom and a suction-gripper as the end-effector, which is controlled by a PIC to carry out the objects via a predefined trajectory to designated locations. A special control algorithm written in C++, is also used to control the servo stepper belt motor and the electro-pneumatic valves of the manipulator. Successful results are evaluated both in assorting of objects in size or property and in synchronous working of the robot arm in connection with image processing.