Corrected components of Zernike Moments for improved content based image retrieval
Pooja Sharma · 2023
In this chapter, we provide a novel solution to image retrieval systems by using corrected real and imaginary coefficients of Zernike Moments (ZMs). ZMs are used as feature vectors representing the global aspect of images. The phase component of ZM is thoroughly investigated and correction are made to it in order to further improve the retrieval rate. Along with that, the histograms of distances between linear edges and centroid of image are used that represent local feature vectors. Both global and local features are analyzed using various similarity measures. Amongst them, the Bray-Curtis similarity measure proved to be better and is used to compute the overall similarity among images. The experimental results reveal that the proposed methods outperform existing recent region and contour based descriptors. The vast analyses also reveal that the proposed system is robust to geometric and photometric transformations. The average retrieval performance over all the databases represents that the proposed (ZMs+HT) attains an 95.49 percent and proposed (ZMs) attains an 89.2 percent accuracy rate.