Recently introduced high-accuracy RGB-D cameras are capable of providing high-quality three-dimension information (color and depth information) easily. The overall shape of the object can be understood by acquiring depth information. However, conventional methods using such cameras use depth information only to extract the local features. To deal with this problem, in our proposed method, the overall object shape is expressed by the depth spatial pyramid based on depth information. Specifically, multiple features within each sub-region of the depth spatial pyramid are pooled. As a result, the feature representation (including the depth topological information) is constructed, where we use the histogram of oriented normal vectors (HONV) as 3D local features and apply locality-constrained linear coding (LLC) to the HONV. Experimental results confirm that the proposed method improves the recognition accuracy compared with conventional methods.

3D-scene-analysis, HONV, LLC, depth-spatial-pyramid.