Advanced Image Retrieval Using Multi-resolution Image Content

We introduce an image retrieval method for searchingan image database by using a query image that has similarity in its global and/or regional image contents to the intended target. Our retrieval method makes use of 2-dimensional wavelet decomposition (2-D WD) of the query and database images. The coefficients of 2-D WD are used as the multi-resolution image contents of the query and database images. The similarity measure between the query and database images is performed after mapping the significant image contents onto the original image. The similarity is estimated by investigating the distance between the positions of the mapped contents and the coincident degree of the operation factors of 2-D WD such as the positive or negative of the coefficients, resolution of decomposition, and frequency information of decomposition. Our method has an advantage to permit a position gap of image contents and a pattern fluctuation, and it is easy to build the image retrieval. 1 Instructions Up...

• Type: Journal
• Year: 2002
• Where: MVA
• Authors: Motohide Yoshimura, Hajime Kiyose, Shigeo Abe
• Keywords: Computer Vision, Image, Image Retrieval, Image Retrieval Method, MVA 2002