Juan Du

Juan Du

Auto-generated profile, based on publicly available information

Juan Du

Department of Basic, Air Force Engineering University, Xi’an, China

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
11 April 2019 in Applied Optics
Verified
Journal Article
Closed access
PDF file
WTWei Tan
Wei Tan
Huixin ZhouHuixin Zhou
Huixin Zhou
JSJiangluqi Song
Jiangluqi Song
HLHuan Li
Huan Li
YYYue Yu
Yue Yu
JDJuan Du
Juan Du
The aim of infrared and visible image fusion is to obtain an integrated image that contains obvious object information and high spatial resolution background information. The integrated image is more conductive for a human or a machine to understand and mine the information contained in the image. To attain this purpose, a fusion algorithm based on multi-level Gaussian curvature filtering (MLGCF) image decomposition is proposed. First, a MLGCF is presented and employed to decompose the input source images into three different layers: small-scale, large-scale, and base layers. Then, three fusion strategies—max-value, integrated, and energy-based—are applied to combine the three types of layers, which are based on the different properties of the three types of layers. Finally, the fusion image is reconstructed by summing the three types of fused layers. Six groups of experiments demonstrate that the proposed algorithm performs effectively in most cases by subjective and objective evaluations and even exceeds many high-level fusion algorithms.

3 January 2020 in Sensors
Verified
Journal Article
Open access
PDF file
JDJuan Du
Juan Du
Huixin ZhouHuixin Zhou
Huixin Zhou
KQKun Qian
Kun Qian
WTWei Tan
Wei Tan
ZZZhe Zhang
Zhe Zhang
YYYue Yu
Yue Yu
Deep learning-based image super-resolution has shown significantly good performance in improving image quality. In this paper, the RGB-IR cross input and sub-pixel upsampling network is proposed to increase the spatial resolution of an Infrared (IR) image by combining it with a color image of higher spatial resolution obtained with a different imaging modality. Specifically, this is accomplished by fusion of the features map of two RGB-IR inputs in the reconstruction of an infrared image. To improve the accuracy of feature extraction, deconvolution is replaced by sub-pixel convolution to upsample image in the network. Then, the guided filter layer is introduced for image denoising of IR images, and it can preserve the image detail. In addition, the experimental dataset, which is collected by us, contains large numbers of RGB images and corresponding IR images with the same scene. Experimental results on our dataset and other datasets demonstrate that the method is superior to existing methods in accuracy and visual improvement.

19 March 2021 in Sensors
Verified
Journal Article
Open access
PDF file
JDJuan Du
Juan Du
KCKuanhong Cheng
Kuanhong Cheng
YYYue Yu
Yue Yu
DWDabao Wang
Dabao Wang
Huixin ZhouHuixin Zhou
Huixin Zhou
Panchromatic (PAN) images contain abundant spatial information that is useful for earth observation, but always suffer from low-resolution ( LR) due to the sensor limitation and large-scale view field. The current super-resolution (SR) methods based on traditional attention mechanism have shown remarkable advantages but remain imperfect to reconstruct the edge details of SR images. To address this problem, an improved SR model which involves the self-attention augmented Wasserstein generative adversarial network ( SAA-WGAN) is designed to dig out the reference information among multiple features for detail enhancement. We use an encoder-decoder network followed by a fully convolutional network (FCN) as the backbone to extract multi-scale features and reconstruct the High-resolution (HR) results. To exploit the relevance between multi-layer feature maps, we first integrate a convolutional block attention module (CBAM) into each skip-connection of the encoder-decoder subnet, generating weighted maps to enhance both channel-wise and spatial-wise feature representation automatically. Besides, considering that the HR results and LR inputs are highly similar in structure, yet cannot be fully reflected in traditional attention mechanism, we, therefore, designed a self augmented attention (SAA) module, where the attention weights are produced dynamically via a similarity function between hidden features; this design allows the network to flexibly adjust the fraction relevance among multi-layer features and keep the long-range inter information, which is helpful to preserve details. In addition, the pixel-wise loss is combined with perceptual and gradient loss to achieve comprehensive supervision. Experiments on benchmark datasets demonstrate that the proposed method outperforms other SR methods in terms of both objective evaluation and visual effect.

26 June 2020 in IEEE Access
Verified
Journal Article
Open access
JDJuan Du
Juan Du
JSJiangluqi Song
Jiangluqi Song
KCKuanhong Cheng
Kuanhong Cheng
ZZZhe Zhang
Zhe Zhang
HZHui-Xin Zhou
Hui-Xin Zhou
HQHanlin Qin
Hanlin Qin
Most CNN-based super-resolution networks require a large number of samples for model training, which may cause overfitting when trained on a specific set, and the internal self-similarity of the test image in this way has also been discarded. To resolve this problem, this paper proposes a novel zero-shot learning (ZSL) network for super resolution. In this model, the training samples are cropped from the input image, which means no extra dataset is required for model training, the overfitting problem in this way can be well avoided. Besides, since all the training samples are cropped from the input itself, the nonlocal self-similarity attributes of the test image can be fully utilized. Moreover, the efficient spatial pyramid of dilated convolutions network with bottleneck (ESP-BNet) is applied in the model as an efficient computational structure to enhance the feature representation. Comparison experiments show that the proposed approach achieves PSNR/SSIM of 31.13dB/87.9% on Set5 dataset and PSNR/SSIM of 27.62dB/78.1% on Set14 dataset, which is around 3dB better than the traditional method and about 1dB higher than those of the state-of-art method.


20 May 2020 in IEEE Access
Verified
Journal Article
Open access
KCKuanhong Cheng
Kuanhong Cheng
JSJiangluqi Song
Jiangluqi Song
JDJuan Du
Juan Du
SRShenghui Rong
Shenghui Rong
HZHuixin Zhou
Huixin Zhou
Single image reflection removal is of great practical importance for various computer vision tasks. Most non-learning methods try to solve this problem through the model-optimization scheme, which fails to produce promising results due to the shortage of suitable priors to model the difference between the reflection layer and the transmission layer. This paper presents an improved generative adversarial network to resolve this problem. First, we suggest that reflection removal is not only a channel-wise separation problem, but also a spatial variational occlusion removal task, which is sensitive to both spatial and channel-wise features. To this end, we integrate the CBAM module into the generator to enhance both spatial and channel-wise feature representation. Second, we consider the reflection layer as a spatial mask with space-relevant reflection intensity information, which can be used to elevate the performance of the discriminator. We then design a novel SNGAN structure with utilize the predicted reflection as a guidance to achieve better adversarial supervision. Specifically, our new generative network has an encoder-decoder structure with skip-connections, where the attention enhancement block is integrated into each skip-connection of the encoder-decoder subnet, and followed by an eight-layer fully convolutional subnet. Furthermore, the SNGAN loss is combined with L2 pixel loss and L1 VGG19 perceptual loss for training. The experimental results with benchmark datasets indicate that our method outperforms several state-of-the-art networks.