期刊
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
卷 14, 期 8, 页码 2793-2813出版社
SPRINGER HEIDELBERG
DOI: 10.1007/s13042-023-01799-5
关键词
Super-resolution; High frequency information; Adversarial generation network; Discrete wavelet transform; Loss function
This article proposes a new image super-resolution reconstruction method called P-WAN, which can improve the utilization of image features from different layers and overcome the problem of blurred texture edges in reconstructed images due to lack of high frequency information. Based on P-WAN, a DP-WAN method is further proposed, which reconstructs better high frequency information and higher quality texture edges by adding discrete wavelet transform. Experimental results demonstrate that the proposed method achieves the best performance in objective evaluation and provides a good visual experience in subjective visual evaluation.
There are some problems with image super-resolution techniques, such as insufficient utilization of image features in different layers and lack high frequency information resulting in blurred texture edges in reconstructed images. A new method for the image super-resolution reconstruction of Widely Activated Network Fused Perceptual Loss (P-WAN) is proposed, which can improve the usage of image features from different layers. The method uses a pre-training model to extract features before activation to obtain the perceptual loss, meanwhile, draws on the adversarial loss in the adversarial generation network, and combines the pixel loss of the image to form a new loss function. Finally, we optimize the loss function by adjusting the weights of the three loss terms. Based on this research, a Widely Activated Network Based on Discrete Wavelet Transform and Fused Perceptual Loss (DP-WAN) is further proposed, which can reconstruct better high frequency information and higher quality texture edges. The method mainly adds the discrete wavelet transform on the basis of P-WAN, trains the different components obtained by the transform separately, and finally reconstructs the super-resolution image through the inverse discrete wavelet transform. To validate the feasibility and effectiveness, 4 representative methods are selected to test on 5 datasets. Experimental results have shown that the proposed method achieves the best performance in objective evaluation, and can obtain a good visual experience in subjective visual evaluation.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据