Color Images To Grayscale Conversions And Its Objective Evaluation Method Research

To save the cost of print, many books, published papers and most newspapersare prone to output in grayscale images; artistic aesthetic derived by the pursuit ofblack and white photography originates at grayscale images; popular research areaslike pattern recognition and computer vision, both chose using small data grayscaleimage to represent the color image quickly during its pre-process. Therefore, theresearch of color-to-gray has important significance and application value. After theproposed of color-to-gray methods, the study of the evaluation of the effect ofcolor-to-gray is developed. But so far methods of evaluation used widely aresubjective evaluation methods. These kind of methods accord to the perception ofhuman,but have expensive time cost and complicate, cannot be combined with otheralgorithms either. Thus the research of the objective evaluation of the effect ofcolor-to-gray is significant and its goal is to achieve objective evaluation score whichconsistent with the subjective evaluation.Based on the key of reproduce the contrast of the original color image are pixelparing and the effect of contrast enhancement, This article improved the Grundlanddecolorization method by proposing a pseudo random pairing method and generatingthe contrast enhancement coefficient adaptively. The improved method extends thegamut of the to-be-converted color image and achieves more vivid contrastenhancement gray images adaptively. Based on the images observed by human notonly reflect the existence of objects, but also embody the psychological factors. Thisarticle introduces the Helmholz-Kohlrausch effect, samples the contrast informationof the original color image and calibrates the dynamic range of grayscale withsaturation to convert color images to gray images. Differences of processing methodof contrast information of color images derived two contrast enhancementcolor-to-gray conversions which accorded with the subjective perception.Experimental results show that both conversions can achieve grayscale images thataccord with the subjective perception and retain most of the information of color image in a short time. Meanwhile, in order to evaluate the effect of the color-to-grayconversions, this article verified the correlation of the results evaluated by theobjective measure indicator and the results of subjective evaluation method, andproposed a measure indicator. Since the merit of judge the performance of theobjective evaluation is comparing with the results of subjective evaluation method,this paper using the multi-linear regression analysis to fit the results of objectiveevaluation to the results of subjective evaluation. Thus the results achieved by theobjective evaluation will closer to the actual visual perception of the eye of human.