Loss Functions for Image Restoration With Neural Networks
Loss Functions for Image Restoration With Neural Networks
复制标题
DOI：
10.1109/tci.2016.2644865
复制
发表时间：
2017-03-01
影响因子：
5.4
通讯作者：
Kautz, Jan
中科院分区：
文献类型：
--
作者：
Zhao, Hang;Gallo, Orazio;Kautz, Jan
神经网络正成为计算机视觉和图像处理领域的核心，不同的体系结构已经被提出来解决特定的问题。然而，在图像处理的背景下，神经网络的损失层的影响并没有得到太多的关注：默认的也是几乎唯一的选择是L(2)。在这篇文章中，我们提出了图像恢复的替代选择。特别是，我们展示了当结果图像要由人类观察者评估时，感知动机损失的重要性。我们比较了几种损失的性能，并提出了一种新的、可微的误差函数。我们表明，即使在保持网络结构不变的情况下，通过更好的损失函数，结果的质量也会显著提高。
Neural networks are becoming central in several areas of computer vision and image processing and different architectures have been proposed to solve specific problems. The impact of the loss layer of neural networks, however, has not received much attention in the context of image processing: the default and virtually only choice is l(2). In this paper, we bring attention to alternative choices for image restoration. In particular, we show the importance of perceptually-motivated losses when the resulting image is to be evaluated by a human observer. We compare the performance of several losses, and propose a novel, differentiable error function. We show that the quality of the results improves significantly with better loss functions, even when the network architecture is left unchanged.