Search

In this paper, we introduce an end-to-end generative adversarial network (GAN) based on sparse learning for single image blind motion deblurring, which we called SL-CycleGAN. For the first time in blind motion deblurring, we propose a sparse ResNet-block as a combination of sparse convolution layers and a trainable spatial pooler k-winner based on HTM (Hierarchical Temporal Memory) to replace non-linearity such as ReLU in the ResNet-block of SL-CycleGAN generators. Furthermore, unlike many state-of-the-art GAN-based motion deblurring methods that treat motion deblurring as a linear end-to-end process, we take our inspiration from the domain-to-domain translation ability of CycleGAN, and we show that image deblurring can be cycle-consistent while achieving the best qualitative results. Finally, we perform extensive experiments on popular image benchmarks both qualitatively and quantitatively and achieve the record-breaking PSNR of 38.087 dB on GoPro dataset, which is 5.377 dB better than the most recent deblurring method., Comment: 12 pages

Wang Y, Tian F, Yan M, et al. Drug Des Devel Ther. 2016;10:1743–1751. At the authors request, the Editor and Publisher of Drug Design, Development and Therapy wish to retract the published article. Previous concerns had been raised to the authors regarding the alleged duplication of images in Figures 1A-b and 1A-c and a corrigendum had been published. More recently, concerns have also been raised to the authors regarding the alleged duplication of images between Figure 2A-b and 2A-e. In addition, there were also concerns raised regarding the duplication of certain features between several images from Figures 2 and 3. The journal wishes to recognize that these particular concerns were introduced during the publication process and were not introduced by the authors. However, in relation to the duplication of images between Figure 2A-b and 2A-e, the authors have acknowledged this. The authors were able to provide original data to the journal, but it was unable to support the integrity of the research and the authors requested to retract the article and the Editor agreed with this. Our decision-making was informed by our policy on publishing ethics and integrity and the COPE guidelines on retraction. The retracted article will remain online to maintain the scholarly record, but it will be digitally watermarked on each page as “Retracted”.

This paper presents a novel context-aware image denoising algorithm that combines an adaptive image smoothing technique and color reduction techniques to remove perturbation from adversarial images. Adaptive image smoothing is achieved using auto-threshold canny edge detection to produce an accurate edge map used to produce a blurred image that preserves more edge features. The proposed algorithm then uses color reduction techniques to reconstruct the image using only a few representative colors. Through this technique, the algorithm can reduce the effects of adversarial perturbations on images. We also discuss experimental data on classification accuracy. Our results showed that the proposed approach reduces adversarial perturbation in adversarial attacks and increases the robustness of the deep convolutional neural network models., Comment: 5 pages, 3 figures, 1 table

The automatic identification and classification of spermatogonium images is a very important issue in biomedical engineering research. This paper proposes a scheme for spermatogonium recognition, in which Zernike moments are used to represent image features. First of all, the mathematical morphology method is employed to extract the intact individual cell in every image, and then we normalize these binary images. Then, Zernike moments are calculated from these normalized images, followed by recognizing the spermatogonia through computing similarity of vectors composed with Zernike moments using Euclidean distance. Experimental results demonstrate that the proposed method, based on Zernike moments, outperforms two well-known methods, namely those based on Hu moments and boundary moments. This method has stronger distinguishing ability, showing better performance in discriminating cell images whether belong to the same cell.

In order to reduce the hardware complexity and unify the sampling rate of transmit filter (TX-filter) module, a novel lookup table (LUT) based zero-multiplication TX-filter structure is proposed for E-band microwave transmission system. Simulation results demonstrate that the proposed structure needs only 16.7% addition operation compared with conventional convolutional based solutions, no multiplexing operation included and an arbitrary sampling rate transform integrated simultaneously.

This paper provides an overview of the current status of e-learning technologies and related theories first. Then present an idea about the integrated access across e-learning platform. On the integrated platform, this paper consider the course as workflow in e-learning scenario and present a strategy about how to adaptive the workflow.