A Fast Codebook Design Algorithm for ECVQ Based on Angular Constraint and Hyperplane Decision Rule(Image)
Year: 2004
Type: article
Abstract: In this paper, we propose two fast codebook generation algorithms for entropy-constrained vector quantization. The first algorithm uses the angular constraint to reduce the search area and to accelerate the search process in the codebook design. It employs the projection angles of the vectors to a reference line. The second algorithm has feature of using a suitable hyperplane to partition the codebook and image data. These algorithms allow significant acceleration in codebook design process. Experimental results are presented on image block data. These results show that our new algorithms perform better than the previously known methods.
Authors: Ahmed Swilem, Kousuke Imamura, Hideo Hashimoto
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W201724189