Improved fast encoding method for vector quantization based on subvector technique

The encoding speed of vector quantization (VQ) is a time bottleneck to its practical applications due to it performing a lot of k-dimensional (k-D) Euclidean distance computations. By using famous statistical features of the sum and the variance of a k-D vector to estimate Euclidean distance first, IEENNS method has been proposed to reject most of unlikely codewords for a certain input vector. By dividing a k-D vector in half to generate its two corresponding (k/2)-D subvectors and then apply IEENNS method again to each of subvectors, SIEENNS method has been proposed as well. SIEENNS method is the so far most search-efficient subvector-based encoding method for VQ but it still has a large memory and computational redundancy. This paper aims at improving state-of-the-art SIEENNS method by introducing a new 3-level data structure to reduce memory redundancy and by avoiding using the variances of two (k/2)-D subvectors to reduce computational redundancy. Experimental results confirmed that the proposed method can reduce memory requirement for each k-D vector from (k+6) to (k+1) and meanwhile improve total search efficiency by 20%̃30% compared to SIEENNS method.