This paper presents a scalable content-based image indexing and retrieval system based on vector wavelet coefficients of color images. Highly decorrelated wavelet coefficient planes are used to acquire a search efficient feature space. The feature space is subsequently indexed using properties of all the images in the database. Therefore, the feature key of an image not only corresponds to the content of the image itself but also to how much the image is different from the other images being stored in the database. The search time linearly depends on the number of images similar to the query image and is independent of the database size. We show that, in a database of 5,000 images, query search takes less than 30 msec on a 266 MHz Pentium II processor, compared to several seconds of retrieval time in the earlier systems proposed in the literature.

This paper presents a scalable approach for content-based searching and browsing of color image archives using segmented color-layout and spatial structure information. The segmented color layout is computed using quantized CIE-Lab* space, and encoded quadtrees have been used to preserve and represent the structural information. Our careful choice of feature space and associated indexing provides a scalable content-based indexing and retrieval (CBIR) system, where query time is relatively independent of the database size and only depends on the number of images similar to the query image. A sample query in our system takes less than 25 msec on a Pentium 233 MHz machine using a database of 5000 512/spl times/512 size images. This query time is orders of magnitude less than the time on available commercial and research prototypes. Experimental results are also presented showing the quality and efficiency of the indexing system.

This paper presents an efficient content based indexing and retrieval mechanism based on vector wavelet coefficients of color images. We use highly decorrelated wavelet coefficient planes to acquire a search efficient feature space. The feature space is subsequently indexed using properties of the all the images in the database. Therefore the feature key of an image does not only correspond to the content of the image itself but also how much the image is different from the other images being stored in the database. The search time depends only on the number of images similar to the query image but not on the size of the entire database. The system is scalable and provides fast retrievals. We show that in a database of 1000 images, query search takes less than 50 msec, on a 266 MHz Pentium processor compared to several seconds of retrieval time in the earlier systems proposed in the literature.