提出了一种基于图像融合的无参考立体图像质量评价算法。该算法利用小波变换分解重构立体图像的左右视图并融合在一幅图像中,归一化处理融合图像的亮度系数,均衡各部分亮度并保留融合图像的结构信息,使用卷积神经网络进行特征提取和回归预测。实验结果表明,所提方法的预测得分与人类主观评价得分具有很好的一致性。

A no-reference stereo image quality assessment algorithm based on image fusion is proposed. The algorithm reconstructs the left and right views of the stereo image by wavelet transform and fuses them into one image. The luminance coefficient of the fused image is normalized, which keeps the brightness of each part in balance and preserves the structural information of the fused image. Finally the convolutional neural network is used to extract feature and predict regression. The experimental results show that the predicted scores of the proposed method are in good agreement with the human subjective assessment scores.