基于感兴趣区域的图像质量评价算法研究

Image Quality Assessment Algorithm Based on Region of Interest

目的降低计算量,提高计算效率,达到评价算法与人眼观察的一致性。方法首先提取感兴趣区域,利用眼动仪记录人眼在观察图像时的关注点、关注时长、眼球运动轨迹以及眼跳数据,然后提取显著区域。将感兴趣区域引入客观视觉模型中评价图像质量,通过测量图像的有效感知跃变百分数,客观评价图像质量的优劣,并且与传统算法进行比较。结果感兴趣区域包含了图像大部分的关键信息,即人眼关注的信息。基于不同增强方法的图像感兴趣区域内的有效感知跃变百分数均大于原始图像的有效感知跃变百分数(20.3%)。结论提出的算法优于传统客观评价算法,且客观质量评价的结果更符合人眼视觉特性。

The aim of this study was to reduce the amount of calculation and improve the computational efficiency, and to achieve the consistency of the assessment algorithm with the human eye observation. First of all, the region of interest was extracted, and the focus, the attention time, the eyeball movement track and the saccade data of human eyes during observation of the image were recorded using an eye tracker. Then the saliency region was extracted. The region of interest was introduced into the objective visual model to evaluate image quality, and the merits of the image quality were objectively evaluated by measuring the percentages of effective perception transitions of the image, and the algorithm of this paper was compared with the traditional algorithm. The region of interest contained most of the key information of the image on which the human eyes focused. The percentages of effective perception transitions of region of interest of image based on different enhancement methods were bigger than that（20.3%） of the original image. The algorithm proposed in this paper was superior to the traditional objective assessment algorithm and the objective quality assessment result accorded more with human visual characteristics.