Reduced-reference quality metrics for measuring the image quality of digitally printed natural images

摘要: The goal of the study was to develop a method for quality computation digitally printed images. We wanted use only attributes which have meaning subjective visual experience Based on the data and our assessments calculation were sharpness, graininess color contrast. proposed metric divides fine detail image into blocks used low energy for graininess calculation. color contrast computes dominant colors using coarse scale image. sharpness coarse scale uses high blocks for reduced reference features metrics are numbers or low blocks. directions colors in overall calculated by combining values. performance proposed application specific compared state art applicationindependent image metric. Linear correlation coefficients between predicted MOS 0.88 for electrophotography 0.98 ink-jet samples, sample set 21 prints electrophotography inkjet, subjectively evaluated 28 observers.