An Image Quality Estimation Model Based on HVS

Based on characteristics of HVS(Human Visual System), we give a model for objective image quality estimation. We use MLR(multiple linear regression) to construct subjective image quality method, MOS(Mean Opinion Score). Pearson and Spearman correlation coefficients of some proposed methods under different kind of distortions, such as compression, noising etc, are analyzed. It's found that each method has different performance. But, however, our constructed quality estimation method is more accurate, monotone and consistent in predicting image quality. What's more, the constructed result is just keep accordance with our objective image quality estimation model. So our model is reasonable and perfect in predicting image quality.