Contrast evaluation can be used as a criterion to evaluate performance of contrast enhancement algorithms and to compare contrast capability of display systems. This paper deals with contrast evaluation models for nat...Contrast evaluation can be used as a criterion to evaluate performance of contrast enhancement algorithms and to compare contrast capability of display systems. This paper deals with contrast evaluation models for natural color images. Two separate models are defined for within- and cross-content evaluations. The former is to differentiate the perceived contrast of the images with the same content. The latter is to discriminate the differences in contrast among the images with different contents. Perception mechanisms are quite different for within- and cross-content evaluations. Local contrast plays an important role in within-content evaluation. In contrast, global contrast dominates the contrast perception for cross-content evaluation. Results of human visual experiments show that the proposed evaluation models outperform previous methods for both within- and cross-content evaluations.展开更多
基金supported by the Inha University Research Grant, Korea
文摘Contrast evaluation can be used as a criterion to evaluate performance of contrast enhancement algorithms and to compare contrast capability of display systems. This paper deals with contrast evaluation models for natural color images. Two separate models are defined for within- and cross-content evaluations. The former is to differentiate the perceived contrast of the images with the same content. The latter is to discriminate the differences in contrast among the images with different contents. Perception mechanisms are quite different for within- and cross-content evaluations. Local contrast plays an important role in within-content evaluation. In contrast, global contrast dominates the contrast perception for cross-content evaluation. Results of human visual experiments show that the proposed evaluation models outperform previous methods for both within- and cross-content evaluations.
