Research On Hardcopy Based Image Quality Evaluation Algorihtm

Nowadays the printing products are flooding our life, no matter the high-quality portrait collection, nor the street leaflets. Both of them need to be output into paper from digital status by printers. In the processing of hardcopies, the decrease of image quality is sometimes unavoidable. Usually we use subjective and objective evaluations as criteria for printer’s quality or user’s satisfaction. However, subjective evaluation is quite time-consuming and uneconomical. Therefore, it is necessary to develop hardcopy-based objective image quality evaluation algorithm.This paper investigates two methods under different conditions. One is based on color patches and the other is based on full reference way. Patch-based way is to test the quality of color patches which are printed with the natural images together. Some noise feathers like mottle and graininess will be detected and built in a model. This method is suitable in non-reference condition and obviously a kind of evolution of traditional image quality assessment algorithm. The full reference way is built based on the former one, but directly measuring the natural image without color patches. Using digital image as reference, the key parts between digital original image and scanned hardcopy are compared and given a quality assessment by human visual algorithm.The two methods have different application fields, but both can reach high prediction accuracy of subjective result by 88%.