This paper proposes a novel feature-similarity (FSIM) index for image quality assessment (IQA), which is based on the idea that the human visual system (HVS) primarily understands images through their low-level features. The FSIM index uses phase congruency (PC), a measure of local structure significance, as the primary feature and gradient magnitude (GM) as the secondary feature. PC and GM are complementary and reflect different aspects of HVS in assessing image quality. The FSIM index is extended to color images as FSIM_C by incorporating chrominance information. Experimental results on benchmark databases show that FSIM and FSIM_C achieve high consistency with subjective evaluations, outperforming other state-of-the-art IQA metrics. FSIM and FSIM_C perform consistently well across all databases, demonstrating their robustness. The FSIM index is computed by combining the similarity measures of PC and GM, weighted by PC values to emphasize significant structural features. The extension to color images involves computing similarity measures for chrominance components and combining them with the luminance-based similarity. The proposed methods show superior performance in terms of correlation with subjective evaluations and root mean squared error, indicating their effectiveness in assessing image quality.This paper proposes a novel feature-similarity (FSIM) index for image quality assessment (IQA), which is based on the idea that the human visual system (HVS) primarily understands images through their low-level features. The FSIM index uses phase congruency (PC), a measure of local structure significance, as the primary feature and gradient magnitude (GM) as the secondary feature. PC and GM are complementary and reflect different aspects of HVS in assessing image quality. The FSIM index is extended to color images as FSIM_C by incorporating chrominance information. Experimental results on benchmark databases show that FSIM and FSIM_C achieve high consistency with subjective evaluations, outperforming other state-of-the-art IQA metrics. FSIM and FSIM_C perform consistently well across all databases, demonstrating their robustness. The FSIM index is computed by combining the similarity measures of PC and GM, weighted by PC values to emphasize significant structural features. The extension to color images involves computing similarity measures for chrominance components and combining them with the luminance-based similarity. The proposed methods show superior performance in terms of correlation with subjective evaluations and root mean squared error, indicating their effectiveness in assessing image quality.