In outdoor low-level vision systems, not only is the resolution of the imaging system important, but rain corrupts the visibility of outdoor scenes and may cause computer vision systems to fail. We present a deep convolutional neural network (CNN) architecture for simultaneously performing single-image super-resolution and rain removal. Instead of learning an end-to-end mapping between the low-resolution rainy images and high-resolution clean images in the original image space, we train our network in the detail space, i.e., the space obtained by high-pass filtering the original image. The proposed CNN has a lightweight structure, yet it outperforms super-resolution and rain removal consecutively by a significantly large margin (>1 dB on average).
This paper presents a hybrid sparse-representation-based approach to single-image super-resolution reconstruction. Our main contribution is threefold: (1) jointly utilize nonlocal similarity of intensity image and low-rank property of gradient image under the framework of sparse representation; (2) incorporate both the high-resolution (HR) and low-resolution dictionaries into the reconstruction process; and (3) incorporate both the unknown HR image and the sparse coefficients into a single objective function. By alternatively minimizing the objective function with respect to the unknown HR image and the sparse coefficients, we get an estimate of the target HR image. Extensive experiments validate that compared with many state-of-the-art algorithms the proposed method yields comparable results for noiseless images and achieves much better results for noisy images.
Low-rank matrix approximation and nonlocal means (NLM) are two popular techniques for image restoration. Although the basic principle for applying these two techniques is the same, i.e., similar image patches are abundant in the image, previously published related algorithms use either low-rank matrix approximation or NLM because they manipulate the information of similar patches in different ways. We propose a method for image restoration by jointly using low-rank matrix approximation and NLM in a unified minimization framework. To improve the accuracy of determining similar patches, we also propose a patch similarity measurement based on curvelet transform. Extensive experiments on image deblurring and compressive sensing image recovery validate that the proposed method achieves better results than many state-of-the-art algorithms in terms of both quantitative measures and visual perception.
Extraction of high-resolution face image is crucial to detect suspect from low-resolution surveillance videos. Though
previously published super-resolution image reconstruction techniques could produce a qualified high-resolution image
from a set of simulated low-resolution images, but the reconstructed image from real low-resolution videos is always
blurring. Two main reasons contribute for this: the process of image registration is ill-posed in nature and the sub-pixel
information provided by the real video sequences is far less sufficient. In this paper, a joint image registration and face
pattern-based high-resolution image reconstruction algorithm was proposed to tackle these two problems. Experimental
results are also provided to demonstrate the effectiveness of the proposed algorithm.
Super-resolution image reconstruction produces a high-resolution image or high- resolution image sequences from a set
of shifted, blurred, and decimated versions thereof, and has been proven to be extremely useful in early vision, video
surveillance, and other applications. However, as magnification increases, previously published techniques get worse
either in computational complexity or ringing artifacts. In this paper, a fast approach is proposed to reduce both the
ringing artifacts and the computational complexity. Experiment results demonstrate that the new approach is more
efficient and can provide much better reconstruction quality in comparison with normal super-resolution algorithms.
KEYWORDS: Reconstruction algorithms, Lawrencium, Image restoration, Signal to noise ratio, Digital filtering, Super resolution, Optical engineering, Detection and tracking algorithms, Algorithm development, Sensors
Superresolution image reconstruction produces a high-resolution image from a set of shifted, blurred, and decimated versions thereof. Previously published techniques perform well only for small magnifications but get worse either in computational complexity or ringing artifacts for large magnifications. We propose a hybrid algorithm to reduce both ringing artifacts and computational complexity of maximum a posteriori (MAP)-based superresolution algorithms. The proposed algorithm magnifies the low-resolution image in stages, and applies a new edge-adaptive postprocessing algorithm at early stages. Experiment results show that the new approach is more efficient and can provide much better reconstruction quality in comparison with normal MAP algorithms.