Soccer Robot Vision System Design For Adaptive Light Intensity

The technology of recognizing by color has been widely used in today's object detection field and images identify field .but this technology is liable to be influenced by asymmetric distribution and instability of the illumination. On the platform of the vision system of robot soccer game in this thesis, the technology of object recognizing and tracing under the condition of asymmetric illustration and changing illustration are investigated.The mainly contents are as follow:Firstly, technical requirement has been done to the self-adaptive vision system. We give the method we will adopt.Secondly, we emphases on some key technologies of the illustration self-adaptive vision system. (1)The color space technique is analyzed. Based on lots of experiments, comparing to the color spaces, we design the system on the basis of YUV color space. And we use a optimize arithmetic of threshold analyze. (2)Aimed at the problem of the asymmetric illustration, we introduce the concept of illumination variety rate, and give an algorithm to calculate it, which combined the concept of histogram distribution. (3)To solve the problem of uneven illumination in time distribution, we utilize the theory of processing of background difference together with a illumination variant vector, and designed a system with the self-adaptive ability.At last, we discuss the design and implement and the subsystem has been tested. We give the frame and expatiate the process, integrate the illustration self-adaptive vision system by combine the key techniques.The result of the system test shows a satisfactory performance in the competitions of robot soccer.