Is the Coordinated Clusters Representation an Analog of the Local Binary Pattern?

Both the Local Binary Pattern (LBP) and the Coordinated Clusters Representation (CCR) are two methods used successfully in the classification and segmentation of images. They look very similar at first sight. In this work we analyze the principles of the two methods and show that the methods are not reducible to each other. Topologically they are as different as a sphere and a torus. In extracting of image features, the LBP uses a specific technique of binarization of images with the local threshold, defined by the central pixel of a local binary pattern of an image. Then, the central pixel is excluded of each local binary pattern. As a consequence, the mathematical basis of the LBP method is more limited than that of the CCR. In particular, the scanning window of the LBP has always an odd dimensions, while the CCR has no this restriction. The CCR uses a binarization as a preprocessing of images, so that a global or a local threshold can be used for that purpose. We show that a classification based on the CCR of images is potentially more versatile, even though the high performance of both methods was demonstrated in various applications.