Skip to Main Content
In the Content-base animation Image Retrieval System, the Opponent Color Local Binary Pattern (OCLBP) algorithm is a typical multi-feature extraction solution, but it has the problems of high dimension of feature and complexity computational. For the defect of Opponent Color Local Binary Patterns, this dissertation presents a new Color-texture feature extraction method. By using the Local Binary Patterns operator in HSV color space, center pixel and neighborhood pixel color feature were extracted from three independent channels, and Local Binary Patterns texture feature was extracted at the same time. Synthesize the color feature vector for a one-dimensional vector, joint the color feature and Local Binary Patterns texture feature. Experiments show that the method is simple calculation and low dimension of animation image feature.
Date of Conference: 29-31 Oct. 2010