摘要
研究表明视觉显著性与图像质量有着密切的关系,图像的失真会导致显著图发生变化。提出了基于显著性的全参考立体图像质量评估方法:利用视差图的前景与背景信息模拟人眼的深度感知特性,且立体显著图被用于计算局部质量,并被作为融合全局质量的加权因子。首先,将深度感知信息与彩色中央眼图的平面显著图融合得到立体图像显著图;然后融合中央眼图的显著性、梯度以及色度得到立体图像的质量得分。实验证明,提出的算法和主观质量得分有较好的一致性。
In this paper,a full-reference stereoscopic image quality assessment method is proposed based on saliency map.The foreground and background information of the disparity map are used to simulate the depth perception characteristics of the human visual system.The three-dimensional saliency maps are utilized to calculate the local image quality,and to fuse the global quality,respectively.The 3 D saliency map of the stereoscopic image is obtained by combining the depth information and the 2 D saliency map,and then the quality score of the stereoscopic image is obtained by merging the visual saliency,gradient and chromaticity of the cyclopean image.
出处
《工业控制计算机》
2018年第10期70-71,74,共3页
Industrial Control Computer
关键词
立体图像
客观评价
视觉显著图
视差
全参考
stereoscopic image
objective assessment
visual saliency
parallax
