A New Biologically Inspired Color Image Descriptor

摘要: We describe a novel framework for the joint processing of color and shape information in natural images. A hierarchical non-linear spatio-chromatic operator yields spatial chromatic opponent channels, which mimics primate visual cortex. extend two popular object recognition systems (i.e., Hmax model sift-based bag-of-words approach) to incorporate along with information. further use combination gist algorithm scene categorization as well Berkeley segmentation algorithm. In all cases, proposed approach is shown outperform standard grayscale/shape-based descriptors alternative schemes on several datasets.