Multi-image super-resolution (MISR) usually outperforms single-image super-resolution (SISR) under a proper inter-image alignment by explicitly exploiting the inter-image correlation. However, the large computational demand encumbers the deployment o
f MISR methods in practice. In this work, we propose a distributed optimization framework based on data parallelism for fast large-scale MISR which supports multi- GPU acceleration, named FL-MISR. Inter-GPU communication for the exchange of local variables and over-lapped regions is enabled to impose a consensus convergence of the distributed task allocated to each GPU node. We have seamlessly integrated FL-MISR into the computed tomography (CT) imaging system by super-resolving multiple projections of the same view acquired by subpixel detector shift. The SR reconstruction is performed on the fly during the CT acquisition such that no additional computation time is introduced. We evaluated FL-MISR quantitatively and qualitatively on multiple objects including aluminium cylindrical phantoms, QRM bar pattern phantoms, and concrete joints. Experiments show that FL-MISR can effectively improve the spatial resolution of CT systems in modulation transfer function (MTF) and visual perception. Besides, comparing to a multi-core CPU implementation, FL-MISR achieves a more than 50x speedup on an off-the-shelf 4-GPU system.
In this paper, we propose a regularization technique for noisy-image super-resolution and image denoising. Total variation (TV) regularization is adopted in many image processing applications to preserve the local smoothness. However, TV prior is pro
ne to oversmoothness, staircasing effect, and contrast losses. Nonlocal TV (NLTV) mitigates the contrast losses by adaptively weighting the smoothness based on the similarity measure of image patches. Although it suppresses the noise effectively in the flat regions, it might leave residual noise surrounding the edges especially when the image is not oversmoothed. To address this problem, we propose the bilateral spectrum weighted total variation (BSWTV). Specially, we apply a locally adaptive shrink coefficient to the image gradients and employ the eigenvalues of the covariance matrix of the weighted image gradients to effectively refine the weighting map and suppress the residual noise. In conjunction with the data fidelity term derived from a mixed Poisson-Gaussian noise model, the objective function is decomposed and solved by the alternating direction method of multipliers (ADMM) algorithm. In order to remove outliers and facilitate the convergence stability, the weighting map is smoothed by a Gaussian filter with an iteratively decreased kernel width and updated in a momentum-based manner in each ADMM iteration. We benchmark our method with the state-of-the-art approaches on the public real-world datasets for super-resolution and image denoising. Experiments show that the proposed method obtains outstanding performance for super-resolution and achieves promising results for denoising on real-world images.
Deformable image registration is a fundamental task in medical imaging. Due to the large computational complexity of deformable registration of volumetric images, conventional iterative methods usually face the tradeoff between the registration accur
acy and the computation time in practice. In order to boost the registration performance in both accuracy and runtime, we propose a fast convolutional neural network. Specially, to efficiently utilize the memory resources and enlarge the model capacity, we adopt additive forwarding instead of channel concatenation and deepen the network in each encoder and decoder stage. To facilitate the learning efficiency, we leverage skip connection within the encoder and decoder stages to enable residual learning and employ an auxiliary loss at the bottom layer with lowest resolution to involve deep supervision. Particularly, the low-resolution auxiliary loss is weighted by an exponentially decayed parameter during the training phase. In conjunction with the main loss in high-resolution grid, a coarse-to-fine learning strategy is achieved. Last but not least, we introduce an auxiliary loss based on the segmentation prior to improve the registration performance in Dice score. Comparing to the auxiliary loss using average Dice score, the proposed multi-label segmentation loss does not induce additional memory cost in the training phase and can be employed on images with arbitrary amount of categories. In the experiments, we show FDRN outperforms the existing state-of-the-art registration methods for brain MR images by resorting to the compact network structure and efficient learning. Besides, FDRN is a generalized framework for image registration which is not confined to a particular type of medical images or anatomy.
