Class-specific codebook construction for biologically inspired recognition

A novel method is presented to improve the object recognition performance of a biologically inspired model by learning class-specific feature codebook. The feature codebook is multi-class shared in the original model, and the content proportion for different codeword type is set in uniform distribution. According to corresponding discriminability, the codebook content proportion is adjusted upon different codeword types (feature vector sizes and filter scales). The test results demonstrate that the codebooks built with proposed modification achieve higher total-length efficiency.