Modified decoded image quality prediction method for fractal image coding

An effective method was proposed to improve the performance of fractal decoded image quality prediction. Firstly, we found that the dynamic range of the linear correlation coefficients (LCCs) between range blocks and their respective best-matched domain blocks was significantly extended by a small number of outliers which contribute little to accumulated collage error and reduced the prediction accuracy. Secondly, to remove the outliers of LCCs, we introduced the effective minimum and maximum of LCCs that can provide the effective bottom and top limits of the actual percentage of accumulated collage error (APACE), EBL-APACE and ETL-APACE, respectively. Finally, by estimating APACE with the average of EBL-APACE and ETL-APACE, the decoded image quality can be directly predicted with the accumulated collage error divided by the estimated APACE. Three state-of-the-art and conventional fractal encoding methods were adopted to verify the effectiveness of the proposed method. Experimental results show that the proposed method can reduce the computational complexity by 2%-7% while maintaining comparable or higher prediction accuracy regarding the previous method.