Average color vector algorithm in color recognition based on a RGB space

In this paper, a novel average color vector algorithm is proposed, which can efficiently resist noise and rapidly segment the different monochromatic targets in a color image. In a feature space, a feature value represents the color vector characteristics of a coordinates point. Calculating color average value and standard deviation and Euclidean distance between a pixel and averages of color vector of trained district, it is possible to judge targets color similarity and recognize rapidly the targets. The tomato simulation experiment founds that it is difficult to recognize two objects of approximate colors. The validity of this method is tested via the simulation experiment by applying average color vector algorithm to the tomato color selecting combining Canny edge detection.