CWSURF: A novel coloured local invariant descriptor based on SURF

Local invariant descriptors is one of the most applicable methods in many applications of image processing (registration, tracking, segmentation, object recognition, etc.). In this domain many useful methods are presented recently such as SIFT, SURF, etc. the main drawback of these methods is their inability to working on colour images. According to valuable information of colour features for construction of better descriptors, some recent efforts are done for improvement of SIFT and SURF using colour information. In this paper we propose a novel colour local invariant descriptor based on SURF (CWSURF) that using a novel colour invariant model (C_w) causes to lower computation cost than colour based SIFT descriptors and higher accuracy (about 5%) than SIFT and SURF.