基于人类视觉系统的立体图像质量评价方法

随着立体图像的大规模发展，很多应用场合需要能够迅速有效地完成对立体图像的质量评价工作，以便于后续应用，而对其进行主观质量评价在效率上很难满足要求。因此，提出了一种感知质量评价算法，并结合了一些人类视觉系统的特性。首先需要得到视差图，然后通过边界图和显著图来对视差图进行加权调整。接着使用Minkowski融合方法将加权后的视差图整合成感知分数。最后，使用多尺度分析来得到最终的感知质量分数。通过使用EPFL立体质量评价数据库来验证文中的立体图像感知质量评价算法。实验显示算法最后得到的客观分数和EPFL数据库中的主观分数具有高度的一致性和单调性，证明了文中的立体图像感知质量评价算法是有效的。

Human Visual System Based Stereoscopic Image Quality Assessment

With the large - scale development of stereoscopic images, many applications need the ability to quickly and efficiently complete the stereoscopic image quality evaluation for further applications, but the subjective quality assessment is difficult to meet the requirements in terms of efficiency. Therefore, the paper proposes a perceptual quality evaluation algorithm, which combines some characteristics of the human visual system. First, the paper obtains the disparity map, then adjusts the disparity map through the weights in the edge map and the saliency map. After that, the Minkowski pooling is used to integrate the weighted disparity map. Finally, the mu]tiscale strategy is applied to compute the final score. The EPFL database is utilized to validate the proposed metric. Experiment shows that the objective score obtained by the proposed metric and the subjective scores have a high degree of consistency and monotonicity. It is proved that the algorithm to evaluate the perceptual quality of stereoscopic images in this paper is very effective.