Abstract: Inspired by the feature extraction during the biologic visual information processing,an algorithm based on a combination of visual perception and local binary pattern histogram Fourier(LBP-HF) is proposed to effectively extract the features to improve the performance of image retrieval.Firstly,principal component map is used as the primary visual feature of intensity.The information of shape and edge is further fused to improve the Itti's model.Therefore,we can obtain 50 feature maps.Secondly,the LBP-HF of each feature maps is computed and concatenated to get an enhanced image feature descriptor vector.Thirdly,the locality preserving projections(LPP) is utilized for dimensionality reduction,and the final low dimensional feature is used for image retrieval.The experimental results show that our method has the discrimination power against color,texture and shape features and has good retrieval performance.