A fast full search equivalent encoding algorithm for image vector quantization based on the WHT and a LUT

Abstract

The application of vector quantization has been constrained to a great deal since its encoding process is very heavy. This paper presents a fast encoding algorithm called the double feature-ordered partial codebook search (DFPS) algorithm for image vector quantization. The DFPS algorithm uses the Walsh-Hadamard transform (WHT) for energy compaction and a look-up table (LUT) for fast reference. The simulation results show that with elaborate preprocessing and memory cost within a feasible level, the proposed DFPS algorithm is faster than other existing search algorithms. Compared with the exhaustive full search (EFS) algorithm, the DFPS algorithm reduces the computational complexity by 97.0% to 97.8% for a codebook size of 256 while maintaining the same encoding quality as that of the EFS algorithm.