Research On The Illumination Robustness Of Vision Systems Of Robocup Middle Size League Soccer Robot

Medium-sized group of robot soccer game is an important project of the robot soccer World Cup (RoboCup),which completely autonomous robot, the most challenging. Robot visual sensor for color target color classification and target recognition, and then complete the path planning, playing, shooting, and other tasks. Medium-sized group of robot soccer game provide a standard test environment.for the research of machine vision technology Research on the illumination robustness of Vision Systems of RoboCup Middle Size League soccer robot can develop machine vision applications, and development of machine vision and image processing theory of knowledge. On the basis of collection and analysis of relevant literature at home and abroad, it mainly aims at off-line color calibration of robot vision and illumination robustness of the color recognition, the main work as followed:(1) the off-line color calibration of robot visionthe offline color calibration is to establish the color lookup table.the color lookup table is a mapping table, which is the color values correspond to the color category. The specific process is the first using the HSI color space that the site image color characteristics; the initial segmentation of the image, ignoring the impact of the luminance component of image segmentation, peak search algorithm of HS two-dimensional color histogram for the initial segmentation; the initial value, the value of the initial segmentation using fuzzy C-means clustering method to compensate for the threshold to generate a color lookup table for real-time game for the classification of the image colors.(2) research on the illumination robustness of the color recognitionThis paper also studied how to improve the robustness of the color recognition under the changing conditions in the illumination. First initialization hue histogram and brightness histogram of the stadium green pixel before game beginning; then use of reverse projection method to find out the green pixel on the count by the hue histogram; Recording the brightness histogram, Compute the lighting rate of change of the brightness histogram of the court in the case of illumination changes, the compensation of illumination changes, update the color lookup table to improve the robustness of color recognition.At the end of this thesis, it summarizes the main research of this task, also points out the shortcomings and the problem which needs further study in the future.