A high-speed codebook design algorithm for ECVQ using angular constraint with search space partitioning [image coding example]

We propose a fast codebook generation algorithm for entropy-constrained vector quantization (ECVQ). The algorithm uses the angular constraint and employs a suitable hyperplane to partition the codebook and image data in order to reduce the search area and accelerate the search process in the codebook design. This algorithm allows significant acceleration in the codebook design process. Experimental results are presented on image block data. These results show that our new algorithm performs better than the previously known methods.