CoMo: a scale and rotation invariant compact composite moment-based descriptor for image retrieval

摘要

Low level features play a significant role in image retrieval. Image moments can effectively represent global information of image content while being invariant under translation, rotation, and scaling. This paper presents CoMo: a moment based composite and compact low-level descriptor that can be used effectively for image retrieval and robot vision tasks. The proposed descriptor is evaluated by employing the Bag-of-Visual-Words representation over various well-known benchmarking image databases. The findings from the experimental evaluation provide strong evidence of high and competitive retrieval performance against various state-of-the-art local descriptors.