Development of a Vision-Based Automation Color Recognition System for Robot Arm

In this study, a machine vision system was developed to recognize difference colors objects in space including red, green, blue colors. The image data from a CMOS camera was analyzed by a PC setup an analysis and image processing software. Besides, to determine position of objects in the world coordinate system, a coordinate calibration method was used. With this method, image coordinate is conversed into the world coordinate and position of objects are accurately determined. The calibration errors were also evaluated. The experiment results show that the proposed color recognition system could detect and identify color and position of target objects correctly. With calibration error, the positive maximum error is 0.56 mm and 0.79 mm in X-direction and Y-direction, respectively. The negative maximum error is −0.54 mm and −0.56 mm in X-direction and Y-direction, respectively. This proposed system was integrated with a self-robot arm with four degrees of freedom to implement classification process.