A new multi-scale deep learning (MDL) framework is proposed and exploited for conducting image interpolation in this paper. The core of the framework is a seeding network that needs to be designed for the targeted task. For image interpolation, a novel attention-aware inception network (AIN) is developed as the seeding network; it has two key stages: 1) feature extraction based on the low-resolution input image; and 2) feature-to-image mapping to enlarge image’s size or resolution. Note that the designed seeding network, AIN, needs to be trained with a matched training dataset at each scale. For that, multi-scale image patches are generated using our proposed pyramid cut, which outperforms the conventional image pyramid method by completely avoiding aliasing issue. After training, the trained AINs are then combined for processing the input image in the testing stage. Extensive experimental simulation results obtained from seven image datasets (comprising 359 images in total) have clearly shown that the proposed MAIN consistently delivers highly accurate interpolated images.

In this paper, a progressive collaborative representation (PCR) framework is proposed that is able to incorporate any existing color image demosaicing method for further boosting its demosaicing performance. Our PCR consists of two phases: (i) offline training and (ii) online refinement. In phase (i), multiple training-and-refining stages will be performed. In each stage, a new dictionary will be established through the learning of a large number of feature-patch pairs, extracted from the demosaicked images of the current stage and their corresponding original full-color images. After training, a projection matrix will be generated and exploited to refine the current demosaicked image. The updated image with improved image quality will be used as the input for the next training-and-refining stage and performed the same processing likewise. At the end of phase (i), all the projection matrices generated as above-mentioned will be exploited in phase (ii) to conduct online demosaicked image refinement of the test image. Extensive simulations conducted on two commonly-used test datasets (i.e., the IMAX and Kodak) for evaluating the demosaicing algorithms have clearly demonstrated that our proposed PCR framework is able to constantly boost the performance of any image demosaicing method we experimented, in terms of the objective and subjective performance evaluations.

The non-local attention mechanism has often been exploited in deep learning to capture long-range dependencies (LRDs) from the same image for enhancing the performance of various image processing methods. However, the initially proposed non-local attention process inevitably yields extremely-high computation complexity, since all the feature points are involved in computing the LRDs. To address this concern, a recently proposed criss-cross network (CCNet), which has a recurrent criss-cross attention (RCCA) module, is used to compute the LRDs by involving only a small set of feature points for significantly reducing computation. Motivated by the RCCA, a novel direction-decoupled non-local attention (DNA) module is proposed in this paper that is able to further reduce the computation complexity of RCCA by half approximately. To verify the performance of our new non-local attention module, a DNA network is developed for conducting single image super-resolution (SISR). Extensive experimental results have clearly demonstrated the superiority of using our DNA network for SISR when compared with that of state-of-the-art methods.

An existing multi-scale residual network (MSRN) has demonstrated its success on conducting the single image super-resolution (SISR) task. The MSRN consists of a number of multi-scale residual blocks (MSRBs), and each MSRB performs convolutions by exploiting two different sizes of windows for conducting multi-scale feature extraction. The smaller window is used to extract image features at a low scale, while the larger one is used for a high scale. To significantly reduce the number of parameters involved in the MSRB, a new feature extraction module, called the asynchronous multi-scale block (AMB), is proposed in this paper. It is based on the fact that the larger window used in the MSRB can be replaced by two smaller windows without affecting the original MSRB's function. Consequently, by replacing each MSRB with our AMB, an asynchronous multi-scale network (AMNet) is then constructed, which can yield a significant reduction on computational complexity. This means that more AMBs can be used in our AMNet to deliver superior SISR performance, while maintaining the same or comparable computational complexity to that of the MSRN. To consolidate all image features generated from all scales, a new fusion scheme, called the adaptive feature fusion block (AFFB), is proposed that weights the extracted features according to their importance for further increasing SISR's performance. Extensive experimental results have clearly shown the superiority of our proposed AMNet when compared with multiple state-of-the-arts.

Curve smoothing has two important applications in computer vision and image processing: 1) the curvature scale-space (CSS) technique for shape analysis, and 2) the Gaussian filter for noise suppression. In this paper, we study how planar curves converge as they are smoothed with increasing scales. First, two types of convergence behavior are clarified. The coined term shrinkage refers to the reduction of arc-length of a smoothed planar curve, which describes the convergence of the curve latitudinally; and another coined term collapse refers to the movement of each point to its limiting position, which describes the convergence of the curve longitudinally. A systematic study on the shrinkage and collapse of three categories of curve models is then presented. The corner models helps to reveal how the local structures of planar curves collapse and what the smoothed curves may converge to. The sawtooth models allows us to gain insights regarding how noise is suppressed from noisy planar curves by the Gaussian filter. Our investigation on the closed curves shows that each curve collapses to a point at its center of mass. However, different curves may yield different limiting shapes at the infinity scale. Finally, based upon the derived results the performance of the CSS technique in corner detection and shape representation is analyzed, and a fast implementation method of the Gaussian filter for noise suppression is proposed.

In aerial images, corner points can be detected to describe the structural information of buildings for city modeling, geo-localization, and so on. For this specific vision task, the existing generic corner detectors perform poorly, as they are incapable of distinguishing corner points on buildings from those on other objects such as trees and shadows. Recently, fully convolutional networks (FCNs) have been developed for semantic image segmentation that are able to recognize a designated kind of object through a training process with a manually labeled dataset. Motivated by this achievement, an FCN-based approach is proposed in the present work to detect building corners in aerial images. First, a DeepLab model comprised of improved FCNs and fully-connected conditional random fields (CRFs) is trained end-to-end for building region segmentation. The segmentation is then further improved by using a morphological opening operation to increase its accuracy. Corner points are finally detected on the contour curves of building regions by using a scale-space detector. Experimental results show that the proposed building corner detection approach achieves an F-measure of 0.83 in the test image set and outperforms a number of state-of-the-art corner detectors by a large margin.