Two new kinds of fast search algorithms for vector quantization

Two new kinds of fast search algorithms are presented for search complexity of vector quantization and other nearest neighbor search problems. The algorithms first find a small area around the input vector with one codevector at least in it, then a hypersphere with one codevector at least is constructed in this small area. These searches are performed by employing two tests, avoiding distortion calculation for those codewords which fail in these tests. Using them, the savings in the number of multiplications can be over 60% in comparison with the full search distance with the partial distance technique, with the number of additions and the number of comparison being reduced dramatically and the coding quality is not degraded. The price is a small increase in memory