期刊名称：International Journal of Computer Science and Network Security
印刷版ISSN：1738-7906
出版年度：2010
卷号：10
期号：1
页码：289-297
出版社：International Journal of Computer Science and Network Security
摘要：Success of image retrieval depends on performance and speed of the retrieval technique used. Many techniques have already been proposed for content based image retrieval, however, the thrust of better performance and faster retrieval is yet present. The paper proposes novel content based image retrieval techniques using row mean, column mean, energy compaction and image splitting. The 32 versions of proposed methods are tested for 55 queries on generic image database of size 1000images spread across 11 classes. With help of precision and recall the performance evaluation of proposed techniques shows that energy compaction gives better quality of retrieved results at much reduced time complexity. Image splitting also helps in improving the performance, more parts gives better retrieval.
Loading...
