结合JND模型的交叉验证深度图质量评价方法

针对在3D视频(3DV)和自由视点视频(FVV)中传统的 图像质量评价方法不适用于深度图的问题,本文从人类视觉感知特性出发,提出一 种新的深度图 质量评估算法。首先进行交叉验证,得到待评价深度图的差值图；然后提取遮挡掩膜,去除 被遮挡的像素 点；再根据人类视觉特性,考虑背景亮度掩蔽、纹理掩蔽和边缘敏感性等因素,应用恰可察 觉失真(JND)模 型得到每个像素点的误差可视阈值；最后计算错误像素率作为度量指标评价深度图的质量 。实验结果表 明,本文提出的算法能够准确地检测错误像素,所提出的度量指标与全参考度量指标的相关 系数的平均值 为0.833,最高达到0.933,与合成虚拟视点均 方误差的相关系数的平均值为0.857,最高达 到0.928。

Depth assessment based on cross validation and JND model

This paper presents a novel depth quality assessment scheme based on human visual perception characteristics.At first,cr oss validation is performed to get the difference map of the depth map to be assessed.Then the oc clusion mask is extracted to remove the occluded pixels.According to the characteristics of hum an vision,considering the factors such as background luminance masking,texture masking and edge sensi tivity,the error visibility threshold of each pixel is obtained by using the model of just notice able distortion (JND).Finall y,the error pixels rate is calculated as a metric to assess the quality of the depth map.Experimental resu lts show that the proposed algorithm can detect the error pixels accurately.The average value of the correlation coefficient between the proposed metric and the full reference metric is 0.8330and the maximum is 0.9338.The average value of the correlation coefficient between the proposed metric and the mean square error of the synthesized virtual view is 0.8579and the maximum is 0.9283.