Image retrieval based on micro-level spatial structure features and content analysis using Full Range Gaussian Markov Random Field model

This paper proposes a novel system, based on Full Range Gaussian Markov Random Field (FRGMRF) model with Bayesian approach for image retrieval. The color image is segmented into various regions according to its structure and nature. The segmented image is modeled to HSV color space, where V attributes to pixel values, and it ranges from 0 to 1. On HSV color space, the texture information and spatial orientation of the pixels are extracted. On each region, the model parameters, autocorrelation coefficient (ACC), and unique texture numbers are computed using the FRGMRF model. The model parameters and ACCs are formed as feature vectors (FVs) of the image. The Mahalanobis distance is applied to measure the distance between the query and target images. Moreover, associated probabilities are computed on the texture numbers on each region, and are used to compute the divergence between the query and target images using cosine function. The obtained results are compared to those of the existing systems. The comparative study reveals that the proposed system outperforms the existing systems.