作者: Chang-Hsing Lee , Ling-Hwei Chen
DOI: 10.1016/0165-1684(95)00009-3
关键词:
摘要: Abstract One of the most serious problems for vector quantization is high computational complexity involved in searching closest codeword through a codebook both design and encoding phases. In this paper, based on assumption that distortion measured by squared Euclidean distance, two high-speed search methods will be proposed to speed up process. The first one uses difference between mean values vectors reduce space. second find Karhunen-Loeve transform (KLT) distribution set training then applies partial elimination method transformed vectors. Experimental results show can lots mathematical operations.