Block DCT Vectors Construction for Face Retrieval Based on Genetic Algorithm

Effective feature extraction is one of the most important steps in content-based face image retrieval applications. A vectors construction method for face retrieval based on block DCT and genetic algorithm is proposed in this paper. According to this method, the block DCT coefficients are firstly selected according to the histogram of each block image, and then genetic algorithm is used to select and construct ultimate discriminative vectors for image retrieval. The experimental results on AR gray face database show that the proposed method is robust to facial situations of pose, expression, illumination and partial occlusion, and can get better performance of retrieval accuracy and speed simultaneously.