Bin Zhang

Dr. Bin Zhang


Research Focus & Metrics

Academic Career Length

18Years (2009-Present)

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

Representative Publications

View All Publications
26 June 2019 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Closed access
BZbin zhang
bin zhang
YZYongjun Zhang
Yongjun Zhang
SWShugen Wang
Shugen Wang
With the growing spatial resolution of satellite images, high spatial resolution (HSR) remote sensing imagery scene classification has become a challenging task due to the highly complex geometrical structures and spatial patterns in HSR imagery. The key issue in scene classification is how to understand the semantic content of the images effectively, and researchers have been looking for ways to improve the process. Convolutional neural networks (CNNs), which have achieved amazing results in natural image classification, were introduced for remote sensing image scene classification. Most of the researches to date have improved the final classification accuracy by merging the features of CNNs. However, the entire models become relatively complex and cannot extract more effective features. To solve this problem, in this paper, we propose a lightweight and effective CNN which is capable of maintaining high accuracy. We use MobileNet V2 as a base network and introduce the dilated convolution and channel attention to extract discriminative features. To improve the performance of the CNN further, we also propose a multidilation pooling module to extract multiscale features. Experiments are performed on six datasets, and the results verify that our method can achieve higher accuracy compared to the current state-of-the-art methods.

2 September 2022 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
BZbin zhang
bin zhang
YZYongjun Zhang
Yongjun Zhang
YLYansheng Li
Yansheng Li
YWYi Wan
Yi Wan
HGHaoyu Guo
Haoyu Guo
KYKun Yang
Kun Yang
Deep convolutional neural networks (CNNs) have gotten a lot of press in the last several years, especially in domains like computer vision (CV) and remote sensing (RS). However, achieving superior performance with deep networks highly depends on a massive amount of accurately labeled training samples. In real-world applications, gathering a large number of labeled samples is time-consuming and labor-intensive, especially for pixel-level data annotation. This dearth of labels in land-cover classification is especially pressing in the RS domain because high-precision, high-quality labeled samples are extremely difficult to acquire, but unlabeled data is readily available. In this study, we offer a new semi-supervised deep semantic labeling framework for semantic segmentation of high-resolution RS images to take advantage of the limited amount of labeled examples and numerous unlabeled samples. Our model uses transformation consistency regularization (TCR) to encourage consistent network predictions under different random transformations or perturbations. We try three different transforms to compute the consistency loss and analyze their performance. Then, we present a deep semi-supervised semantic labeling technique by using a hybrid transformation consistency regularization (HTCR). A weighted sum of losses, which contains a supervised term computed on labeled samples and an unsupervised regularization term computed on unlabeled data, may be used to update the network parameters in our technique. Our comprehensive experiments on two RS datasets confirmed that the suggested approach utilized latent information from unlabeled samples to obtain more precise predictions and outperformed existing semi-supervised algorithms in terms of performance. Our experiments further demonstrated that our semi-supervised semantic labeling strategy has the potential to partially tackle the problem of limited labeled samples for high-resolution RS image land-cover segmentation.

17 December 2022 in Journal of Clinical Medicine
Verified
Journal Article
Open access
PDF file
GLGuihua Liu
Guihua Liu
BLBingqin Li
Bingqin Li
HRHua Rong
Hua Rong
BDBEI DU
BEI DU
BWBiying Wang
Biying Wang
RWRuihua Wei
Ruihua Wei
This study aimed to explore the effect of repeated low-level red light (RLRL) on axial length (AL), choroid blood flow, and anterior segment components in myopic adults. Ninety-eight myopic adults were randomly divided into the RLRL group (n = 52) and the control group (n = 46). Subjects in the RLRL group completed a 4-week treatment composed of a 3-min RLRL treatment session twice daily, with an interval of at least 4 h. Visits were scheduled before and on 7, 14, 21, and 28 days after the treatment. AL, subfoveal choroidal thickness (SChT), choroidal vascularity index (CVI), and anterior segment parameters were measured at each visit. A linear mixed-effects model showed that the AL of the subjects in RLRL decreased from 24.63 ± 1.04 mm to 24.57 ± 1.04 mm, and the SChT thickened by 18.34 μm. CVI had a slight but significant increase in the 0–6 zone. However, all the anterior segment parameters did not change after RLRL treatment. Our study showed that the choroid’s thickening is insufficient to explain the axial length shortening. The unchanged anterior segment and improved choroid blood flow suggest that the AL shortening in this study is mainly related to changes in the posterior segment.

1 November 2018 in Translational Vision Science & Technology
Verified
Journal Article
Open access
PDF file
JWJinghui Wang
Jinghui Wang
DYDan Yang
Dan Yang
HBHua Bi
Hua Bi
BDBei Du
Bei Du
WLWeiping Lin
Weiping Lin
RWRuihua Wei
Ruihua Wei
Purpose: We present a new method for analyzing relative corneal refractive power (RCRP) in children undergoing orthokeratology and explore its potential association to effective myopic control Methods: A total of 55 children aged 8 to 12 years participated in the study. Axial growth was calculated as the difference in axial length before and 1 year after orthokeratology. Growth <0.30 mm was considered as effective control. Corneal topography was obtained before and 4 months after lens dispatch. The topography was divided into 36 10° slices and the maximal RCRP (mRCRP) in each was calculated and fitted into a model that integrated the effects of mean refractive power (M), corneal asymmetry (f1), and astigmatism (f2). The relationship between the probability of achieving effective control and the modulation of mRCRP was analyzed with logistic regression. Results: A total of 45 subjects achieved effective control, but for 10 the treatment was ineffective. The M-values were not different between the groups. Modulations of mRCRP were significantly larger in the effective than the ineffective group (1.17 vs. 0.64 diopters [D] for f1, P = 0.02; 0.85 vs. 0.35 D for f2, P = 0.03). The probability to achieve effective control increased with modulation of mRCRP (P = 0.02). With a peak mRCRP > 4.5 D, a subject had an above 80% chance to achieve effective control. Conclusions: The new method reveals that how the combination of spherical equivalent (SE), corneal asymmetry, and astigmatism determines modulation of the mRCRP and a large amplitude of modulation is associated with a higher probability of effective myopic control. Translational Relevance: Our finding enables clinicians to estimate the outcome early and provides new insights to lens design.

29 December 2022 in IEEE Geoscience and Remote Sensing Letters
Verified
Journal Article
Closed access
BZbin zhang
bin zhang
YZYongjun Zhang
Yongjun Zhang
YLYansheng Li
Yansheng Li
YWYi Wan
Yi Wan
YYYongxiang Yao
Yongxiang Yao
Clouds inevitably exist in satellite images, which limits the processing and application of satellite images to a certain extent. Therefore, cloud detection is a preprocessing task in satellite image extraction and analysis processing. However, the existing methods are difficult to mine robust features, and the number of parameters and computation is large, which is not conducive to the deployment of the model. In this letter, CloudViT, a lightweight vision transformer network for cloud detection from satellite imagery, is proposed. In detail, to utilize dark channel priors in multispectral imagery to guide the network to learn features, a multi-scale dark channel extractor is used to first predict dark channels, and then the dark channel features and image features are input to the attention mechanism-based dark channel-guided context aggregation module to enhance image features, which in turn makes cloud detection results more accurate. At the same time, to enhance the transfer ability of the network between different satellite sensors, a plug-and-play channel adaptive module is proposed to deal with the inconsistency of the number of different satellite sensor bands. Experimental results on the Landsat7 dataset show that our network CloudViT outperforms state-of-the-art methods while keeping the number of parameters and computation small. At the same time, the experimental results on transfer to three other datasets show that using the channel adaptation module can greatly improve the transfer ability of the model.

18 September 2020 in Current Eye Research
Verified
Journal Article
Closed access
XYXiaoyan Yang
Xiaoyan Yang
HBHua Bi
Hua Bi
LLLihua Li
Lihua Li
SLShumao Li
Shumao Li
SCSong Chen
Song Chen
YWYan Wang
Yan Wang
To quantify the spatial distribution of relative corneal refractive power shift (RCRPS) to investigate its association with axial length growth. Eighty myopic children were randomly assigned for fi...