A color-based scale-invariant interest points matching approach

The scale-invariant interest points matching approaches always require a lot of computation and sometimes fail to meet the real-time demands in some application fields. To solve the problem, a new local scale-invariant descriptor based on the image color information is proposed in this paper. The descriptor consists of concentric circle structure which is divided into several subregions. The radiuses of the concentric circles are proportional to the scale factor, and the coordinates of the descriptor rotate in relation to the interest point orientation. Meanwhile, the descriptor chooses the mean values of different color components R, G, B in each subregion as the feature vector´s elements to reduce the descriptor´s dimension. Compared with other descriptors, this descriptor is robust to many image transformations and is indicative of less computation. The performance of the approach is analyzed in theory in this paper and the experimental results have certified its validity too.