年間 12 号発行
ISSN 印刷: 0040-2508
ISSN オンライン: 1943-6009
Indexed in
Entropy-Like Measure of Background Content for Image Retrieval and Sorting in Large Databases
要約
An approach to analysis of properties of color RGB images based on analysis of entropy-like measure for each component is put forward. It is demonstrated that the proposed measure can be effectively used for indexing, search and retrieval such images from large databases that contain a large percentage (sufficient amount) of pixels belonging to homogeneous color background. This allows indexing and sorting images according to this feature and exploiting this information for retrieval images from large size databases. The proposed measure can be also useful for further processing of images since it might produce useful information for strategy of image feature determination. The results of testing the entropy-like measure in processing of Yandex (www.yandex.ru) database of image thumbnails are also presented.