Speckle noise limits the usage of synthetic aperture radar (SAR) for object recognition and segmentation tasks. Most traditional methods sacrifice useful image information to achieve speckle reduction. The classic method based on local sliding window filtering has obvious side effect of erasing object edges and blurring texture information comparing with ground truth image. Another widely used method is convolutional neural network based on mean squared error, the visual effect of denoised image is not satisfactory even though MSE loss can have higher peak signal-to-noise ratio (PSNR) performance. In this paper, we present a cascade network to address this problem, namely SRNet, which employs an asymmetric architecture for the task of speckle noise reduction. The cascade architecture can supervise the network to revise on both pixel-wise level and feature-wise level by calculating correlation coefficient loss on the feature maps. In the meanwhile, we utilize the auxiliary loss on the intermediate results to accelerate the convergence of the network. The proposed network preserves the edge texture details much better than other compared methods.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.