Abstract
In this paper, we present a new two-dimensional (2D) histogram technique for colour image segmentation using HSV (hue, saturation, and value) colour space. The approach is achieved into two main steps. In the first one, we build a 2D histogram according to a predefined threshold function derived from properties of HSV and the change of the pixel values in a block of 3 × 3 neighbouring pixels. In the second one, we use a hierarchical analysis of 2D histogram for detecting the classes, and we perform the segmentation algorithm by reclassification of the pixels not classified in the determined classes, according to Euclidian distance. The method has been applied on synthetic images and reel images devoted to agriculture application. This segmentation technique gives promising results compared with those obtained either by 1D histogram using HSV or 2D histogram using RGB space.