Color-Shape Context for Object Recognition
Aristeidis Diplaros, Student Member, Theo Gevers, Ioannis Patras · UvA-DARE (University of Amsterdam) · 2003
In this paper, we study computational models and techniques to merge color and shape invariant information to recognize objects. We propose a feature, which we call color shape context, and it is a histogram that combines the spatial (shape) and color information of the image in one compact representation. This histogram codes the locality of color transitions in an image. Illumination invariant derivatives are first computed and provide the edges of the image, which is the shape information of our feature. These edges are used to obtain similarity (rigid) invariant shape descriptors. The color transitions that take place on the edges are coded in an illumination invariant way and are used as the color information. The color and shape information are combined in one multidimensional vector. Our experiments show that the feature is invariant to the similarity transformations of shape such as translation, rotation and scaling and also to noise and illumination changes of color. We conducted our experiments in three databases whose size ranges from 500 to 7200 images. We report considerably better results than only color-based or only shape-based methods. We also found experimentally that the feature exhibits robustness to viewpoint changes for the COIL-100 dataset.