![]() The PCA performs an orthogonal transformation to convert a set of potentially correlated variables into a set of variables that are not linearly correlated. The LBG algorithm finds a codebook based on the better vectors sent to an initial codebook by the PCA. The PCA-LBG-based algorithms include (1) PCA-LBG-Median, which selects the median vector of each group, (2) PCA-LBG-Centroid, which adopts the centroid vector of each group, and (3) PCA-LBG-Random, which randomly selects a vector of each group. All training vectors are grouped according to the projected values of the principal components. Vector quantisation (VQ) codebooks are generated by combining principal component analysis ( PCA) algorithms with Linde-Buzo-Gray (LBG) algorithms. PCA-LBG-based algorithms for VQ codebook generation
0 Comments
Leave a Reply. |