REDUCED REFERENCE IMAGE QUALITY ASSESSMENT USING WAVELET COEFFICIENT CO-OCCURRENCE MATRIX

Image quality is a key characteristic while sharing images through social media platforms. Reduced-reference Image Quality Assessment (RRIQA) evaluates image quality degradation. This method extracts partial information from reference and obtains high quality index that is consistent with human perception. In this paper, we propose an RRIQA algorithm based on Wavelet–Frequency Co-occurrence Matrix (WFCoM). In proposed algorithm single level decomposition of grey level image leads to approximate and detail coefficients. By applying DFT to coefficient matrix, we get Fourier spectrum of the image. By adjusting the offset as in GLCM, we have identified the relationship between frequency points in the frequency spectrum. Feature vectors such as Entropy, Correlation, and Homogeneity are extracted from reference and distorted image. The results on Live database validate that the WFCoM outperforms other objective models w.r.t small number of features and reduction in data rate.