Abstract
Image retrieval procedures locate interest points and consider the features as the visual property of an image in computer vision. These primitive features define visual attributes as local or global features for content based image retrieval. The visual attributes of an image, include spatial information, shape, texture, object and color, describe the image category. The present research contributes feature detector by performing non-max suppression after detecting edges and corners based on corner score and pixel derivation-based shapes on intensity-based interest points. Thereafter, interest point description applied on interest point features set by applying symmetric sampling to cascade matching produced by validating dense distributed receptive fields after estimating perifoveal receptive fields. Spatial color-based features vector are fused with retinal and color-based feature vector extracted after applying L2 normalization on spatially arranged color image. Dimensions are reduced by applying PCA on massive feature vectors produced after symmetric sampling and transmitted to bag-of-word in fused form for indexing and retrieval of images. Extensive experiments are performed on well-known benchmarks corel-1000, core-10000, caltech-101, image net, alot, coil, ftvl, 102-flowers and 17-flowers. In order to measure the competitiveness, we designed a comparison of proposed method with seven descriptors and detectors RGBLBP, LBP, surf, sift, DoG, HoG and MSER. The proposed method reports remarkable AP, AR, ARP, ARR, P&R, mAP and mAR rates in many categories of image datasets.