Search

The dominance of visual data in recent times has made a fundamental change to our everyday life. Less than five to ten years ago, Internet and World Wide Web were not the daily vocabulary for the general public. But now, even a young child can use the Internet to search for information. This, however, does not mean that we have a mature technology to perform visual information search. On the contrary, visual information retrieval is still in its infancy. The problem lies on the semantic richness and complexity of visual information in comparison to alphanumeric information. In this thesis, we present new paradigms for content-based image indexing and retrieval for Visual Information Systems. The concept of Image Hashing and the developments of Composite Bitplane Signatures with Inverted Image Indexing and Compression are the main contributions to this dissertation. These paradigms are analogous to the signature-based indexing and inversion-based postings for text information retrieval. We formulate the problem of image retrieval as a two dimensional hashing as oppose to a one-dimensional hash vector used in conventional hashing techniques. Wavelets are used to generate the bitplane signatures. The natural consequence to our bitplane signature scheme is the superimposed bitplane signatures for efficient retrieval. Composite bitplanes can then be used as the low-level feature information together with high-level semantic indexing to form a unified and integrated framework in our inverted model for content-based image retrieval.

The purpose of this investigation was to determine the color preference of Thai and American students for text and background computer color combinations. The primary purpose of this study was to determine if there were differences between Thai and American students' computer color combination preferences.