Single Image Super-Resolution with Arbitrary Magnification Based on High-Frequency Attention Network

Among various developments in the field of computer vision, single image super-resolution of images is one of the most essential tasks. However, compared to the integer magnification model for super-resolution, research on arbitrary magnification has been overlooked. In addition, the importance of single image super-resolution at arbitrary magnification is emphasized for tasks such as object recognition and satellite image magnification. In this study, we propose a model that performs arbitrary magnification while retaining the advantages of integer magnification. The proposed model extends the integer magnification image to the target magnification in the discrete cosine transform (DCT) spectral domain. The broadening of the DCT spectral domain results in a lack of high-frequency components. To solve this problem, we propose a high-frequency attention network for arbitrary magnification so that high-frequency information can be restored. In addition, only high-frequency components are extracted from the image with a mask generated by a hyperparameter in the DCT domain. Therefore, the high-frequency components that have a substantial impact on image quality are recovered by this procedure. The proposed framework achieves the performance of an integer magnification and correctly retrieves the high-frequency components lost between the arbitrary magnifications. We experimentally validated our model’s superiority over state-of-the-art models.

TWIST-GAN: Towards Wavelet Transform and Transferred GAN for Spatio-Temporal Single Image Super Resolution

Single Image Super-resolution (SISR) produces high-resolution images with fine spatial resolutions from a remotely sensed image with low spatial resolution. Recently, deep learning and generative adversarial networks (GANs) have made breakthroughs for the challenging task of single image super-resolution (SISR) . However, the generated image still suffers from undesirable artifacts such as the absence of texture-feature representation and high-frequency information. We propose a frequency domain-based spatio-temporal remote sensing single image super-resolution technique to reconstruct the HR image combined with generative adversarial networks (GANs) on various frequency bands (TWIST-GAN). We have introduced a new method incorporating Wavelet Transform (WT) characteristics and transferred generative adversarial network. The LR image has been split into various frequency bands by using the WT, whereas the transfer generative adversarial network predicts high-frequency components via a proposed architecture. Finally, the inverse transfer of wavelets produces a reconstructed image with super-resolution. The model is first trained on an external DIV2 K dataset and validated with the UC Merced Landsat remote sensing dataset and Set14 with each image size of 256 × 256. Following that, transferred GANs are used to process spatio-temporal remote sensing images in order to minimize computation cost differences and improve texture information. The findings are compared qualitatively and qualitatively with the current state-of-art approaches. In addition, we saved about 43% of the GPU memory during training and accelerated the execution of our simplified version by eliminating batch normalization layers.

Lightweight Multi-Scale Asymmetric Attention Network for Image Super-Resolution

Recently, with the development of convolutional neural networks, single-image super-resolution (SISR) has achieved better performance. However, the practical application of image super-resolution is limited by a large number of parameters and calculations. In this work, we present a lightweight multi-scale asymmetric attention network (MAAN), which consists of a coarse-grained feature block (CFB), fine-grained feature blocks (FFBs), and a reconstruction block (RB). MAAN adopts multiple paths to facilitate information flow and accomplish a better balance of performance and parameters. Specifically, the FFB applies a multi-scale attention residual block (MARB) to capture richer features by exploiting the pixel-to-pixel correlation feature. The asymmetric multi-weights attention blocks (AMABs) in MARB are designed to obtain the attention maps for improving SISR efficiency and readiness. Extensive experimental results show that our method has comparable performance with fewer parameters than the current advanced lightweight SISR.

Single-image super-resolution improvement of X-ray single-particle diffraction images using convolutional neural network

Femtosecond X-ray pulse lasers are promising probes for elucidating the multi-conformational states of biomolecules because they enable snapshots of single biomolecules to be observed as coherent diffraction images. Multi-image processing using an X-ray free electron laser has proven to be a successful structural analysis method for viruses. However, some difficulties remain in single-particle analysis (SPA) for flexible biomolecules with sizes of 100 nm or less. Owing to the multi-conformational states of biomolecules and the noisy character of diffraction images, diffraction image improvement by multi-image processing is not always effective for such molecules. Here, a single-image super-resolution (SR) model was constructed using a SR convolutional neural network (SRCNN). Data preparation was performed in silico to consider the actual observation situation with unknown molecular orientations, and fluctuation of molecular structure and incident X-ray intensity. It was demonstrated that the trained SRCNN model improved the single-particle diffraction image quality, which corresponded to an observed image with an incident X-ray intensity; i.e., approximately three to seven times higher than the original X-ray intensity, while retaining the individuality of the diffraction images. The feasibility of SPA for flexible biomolecules with sizes of 100 nm or less was dramatically increased by introducing the SRCNN improvement at the beginning of the variety structural analysis schemes.

BLIND RESTORATION USING CONVOLUTION NEURAL NETWORK

Image restoration is a branch of image processing that involves a mathematical deterioration and restoration model to restore an original image from a degraded image. This research aims to restore blurred images that have been corrupted by a known or unknown degradation function. Image restoration approaches can be classified into 2 groups based on degradation feature knowledge: blind and non-blind techniques. In our research, we adopt the type of blind algorithm. A deep learning method (SR) has been proposed for single image super-resolution. This approach can directly learn an end-to-end mapping between low-resolution images and high-resolution images. The mapping is expressed by a deep convolutional neural network (CNN). The proposed restoration system must overcome and deal with the challenges that the degraded images have unknown kernel blur, to deblur degraded images as an estimation from original images with a minimum rate of error.