Soccer field detection based onhistogram of s-rgb

Field detection and extraction are the first task in the object recognition in soccer robot. In this paper we propose a novel method of background detection. It based on color appearance. In this method, we assume that the background has a dominant color in a frame. First, we calculate the sum of RGB intensity (we called as s-RGB) in each pixel. Then we group s-RGB in 16 bins to form a histogram of s-RGB. Second, we obtain a modus of s-RGB. This modus is used to determine the interest bin. Third, from this interest bin, we create histogram of R, G and B intensity, then we calculate modus of R, G and B intensity. These intensities are supposed to be the background color. Applying a threshold, we can extract foreground. For image size of 640×480 pixels, the computation time is 97.29 ms, suitable for real time application.