New image pre-processing method based on linear transformation of YUV color space

In the field of robot vision using color segmentation, selecting a proper threshold is usually a difficult issue. This paper makes a statistical analysis to the images collected in RoboCup 4-legged league tournament, and the result shows that the YUV color space is not fully used. So an image pre-processing method using linear transformation of the YUV space is introduced. This method can makes the significant colors with information expanding to the entire YUV color space and consequently reinforce the boundaries between different colors, which can, to a certain extent, reduce the difficulty of selecting threshold for color segmentation.This paper also introduces the color calibration table approach, making robot real-time computing costs greatly reduced.