Object of research - processes of adaptive vector quantization, palette encoding and content-based image retrieval for customized television systems. Subject of research - systems for adaptive vector quantization, palette encoding and content-based image retrieval for customized television systems. Research methods - methods of linear algebra, mathematical statistics, theory of algorithms, mathematical and computer simulation were used to develop the methods of adaptive vector quantization, palette encoding and content-based image retrieval. Theoretical and practical results: the method of adaptive vector quantization of TV images is first proposed, in this method, in contrast with the existing methods, the number of quantization levels is determined by the number of grayscale levels, it allows to encode images using the uniform code with adaptive length and 217 times on average reduce the color redundancy while providing the sufficient quality of visual perception with 39.13 dB PSNR on average; the method for comparing the low-level color descriptors is first suggested, this method, in contrast with the existing methods, provides the order relation of search results and allows to range the search results in relevance descending order; the method for adaptive palette encoding of video was improved, in this method, in contrast with the existing methods, nearly 3 times reduce of the file size is achieved while providing the sufficient quality of visual perception; the method of extraction of low level color descriptor was improved, this method, in contrast with the existing methods, has small computational complexity linearly dependent on the image size and ensures interoperability of search engines; the codec with an adaptive palette and 1 byte code length was implemented, this codec supports the BMP format and allows to view the image using standard image editors; the single-byte palette image format VBMP with adaptive code length and the appropriate codec were developed, these codec and format can be used for encoding, storing, processing and transmission of medical and synthetic television images in customized television systems; the single-byte palette video format and video codec were implemented, these codec and format are appropriate for the storage and transmission of video sequences in customized television systems; the software for the adaptive vector quantization of images was developed, it can be used in customized television systems for pre-processing and transmission of images; the software for extracting and comparing the low-level color descriptors was developed, it can be integrated into existing content-based image retrieval systems.
