An object recognition method is proposed for cluttered color and depth scene, which is based on color reach histograms. The motivation of the research project is came from the difficulty in searching colored objects in cluttered scene with respect to color and depth. We propose color reach histograms (CRH) for feature representation in order to combine color frequency and depth distribution in the scene. The proposed method can be applied to searching in both 2D and 3D scene. By using CRH, in comparison with conventional histogram-based methods, the stability and robustness for illumination change and occlusion could be obtained and verified through many experiments with real objects.
展开▼
