As a guest user you are not logged in or recognized by your IP address. You have
access to the Front Matter, Abstracts, Author Index, Subject Index and the full
text of Open Access publications.
Most of the super-resolution methods based on deep neural networks use a fixed image degradation model. However, when the real image degradation is inconsistent with the model assumptions, the performance of the network will be severely reduced. To address this issue, we propose a degradation-oriented adaptive network (DOANet), which can be used to solve multiple degradation super-resolution reconstruction. The network includes a degradation estimation network and an adaptive reconstruction network. The degradation estimation network extracts the abstract degradation of LR images using continuous residual blocks to provide necessary degradation information for reconstruction network. Then, the adaptive reconstruction network dynamically adjusts the parameters of the network layers based on the estimated degradation information using dynamic convolution and channel modulation to deal with various input images. Experimental results show that the proposed DOANet can cope well with multiple degradation and is superior to other recent blind methods in qualitative and quantitative comparison.
This website uses cookies
We use cookies to provide you with the best possible experience. They also allow us to analyze user behavior in order to constantly improve the website for you. Info about the privacy policy of IOS Press.
This website uses cookies
We use cookies to provide you with the best possible experience. They also allow us to analyze user behavior in order to constantly improve the website for you. Info about the privacy policy of IOS Press.