Lightweight models are currently the focal point in image super-resolution (ISR) research, of which the application on resource-limited devices is constrained by heavy computational requirements. As an efficient approach to enhance the inference efficiency of deep learning models, low-bit quantization has garnered significant interest. In this paper, we emphasize that low-bit ISR is not merely a parody of its full-precision version and explore binary quantization in ISR from the perspective of information transmission, pushing the limits of binarized ISR. Specifically, we propose a Maximum Entropy Routing (MER) mechanism to dynamically control activation distribution, maximizing the information entropy of binarized feature maps. Additionally, a Learnable Deviation Compensation (LDC) and an Adaptive Step-size Estimation (ASE) are introduced to reduce information loss during the forward and backward passes, respectively. By enabling smoother information transmission through more flexible binarized activation representations and more precise gradient estimation, the performance gap between binarized and full-precision models is narrowed to less than 0.3 dB. Extensive experiments demonstrate that our proposed binarization method achieves state-of-the-art results in Peak Signal-to-Noise Ratio (PSNR) across all popular benchmarks.

Laser speckle contrast imaging (LSCI) is a wide-field and noncontact imaging technology for mapping blood flow. Although the denoising method based on block-matching and three-dimensional transform-domain collaborative filtering (BM3D) was proposed to improve its signal-to-noise ratio (SNR) significantly, the processing time makes it difficult to realize real-time denoising. Furthermore, it is still difficult to obtain an acceptable level of SNR with a few raw speckle images given the presence of significant noise and artifacts. A feed-forward denoising convolutional neural network (DnCNN) achieves state-of-the-art performance in denoising nature images and is efficiently accelerated by GPU. However, it performs poorly in learning with original speckle contrast images of LSCI owing to the inhomogeneous noise distribution. Therefore, we propose training DnCNN for LSCI in a log-transformed domain to improve training accuracy and it achieves an improvement of 5.13 dB in the peak signal-to-noise ratio (PSNR). To decrease the inference time and improve denoising performance, we further propose a dilated deep residual learning network with skip connections (DRSNet). The image-quality evaluations of DRSNet with five raw speckle images outperform that of spatially average denoising with 20 raw speckle images. DRSNet takes 35 ms (i.e., 28 frames per second) for denoising a blood flow image with 486×648 pixels on an NVIDIA 1070 GPU, which is approximately 2.5 times faster than DnCNN. In the test sets, DRSNet also improves 0.15 dB in the PSNR than that of DnCNN. The proposed network shows good potential in real-time monitoring of blood flow for biomedical applications.

Laser speckle contrast imaging (LSCI) is a full-field, noncontact imaging technology for mapping blood flow with high spatio-temporal resolution, in which the speckle contrast can be estimated either in spatial domain or temporal domain. Temporal LSCI (tLSCI) provides higher spatial resolution than spatial domain does. However, when the number of sampling frames is limited, it is difficult to obtain accurate blood flow velocity owing to the significant statistical noise. The widely used spatially averaged tLSCI (savg-tLSCI) usually requires a large number of sampling frames to obtain acceptable denoising performance. Here, based on the nonlocal filtering strategy of block-matching and three-dimensional transform-domain collaborative filtering (BM3D), Manhattan distance-based adaptive BM3D (MD-ABM3D) is proposed to effectively manage the complicated inhomogeneous noise in tLSCI image and improve the signal-to-noise ratio. Manhattan distance improves the accuracy of the block matching in strong noise, and the adaptive algorithm adapts to the inhomogeneous noise and estimates suitable parameters for improved denoising. MD-ABM3D improves 4.91 dB in peak signal-to-noise ratio relative to savg-tLSCI. It achieves stability for denoising tLSCI image with different temporal windows. The image-quality evaluation of MD-ABM3D for tLSCI (t = 20 frames) equals that of savg-tLSCI (t = 60 frames). It achieves high signal-to-noise ratio with a reduced number of sampling frames. A reduced number of sampling frames are more practical for biomedical applications. It also offers higher temporal resolution and less disturbance from the motion of the moving object.