An Autonomous Robot Oriented Real-Time Color Vision System

A set of realtime identification software system was developed which is based on a color segmentation as the core of the whole system. The system is composed of a plug and play USB digital camera. First, the color space of input color image is converted from RGB color space to YIQ color space. So the information of IQ plane is obtained. Second, the image in IQ plane is segmented by the given color. After that, a run length encoded (RLE) version is computed. Then the information of the image is extracted from the RLE version. At last, the vision information is transferred to the robot control system through the TCP/IP interface. In order to restrain the influence of the variational intensity of the light, a threshold learning method was adopted. It is shown by the experiments and practices that the whole system has good realtime performance and satisfying identification effect.