Exemplar-Based Denoising: A Unified Low-Rank Recovery Framework

Zhang, Xiaoqin; Zheng, Jingjing; Wang, Di; Zhao, Li

doi:10.1109/tcsvt.2019.2927603

Cited by 43 publications

(19 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, this problem can be overcome in subsequent work by introducing parallel computing techniques and high arithmetic devices. The superior optimization performance of EACOR ensures efficient image segmentation models, and provide greater possibilities for application to other fields in the future, such as disease prediction ( Su et al, 2019 ; Li L. et al, 2021 ), recommender system ( Li et al, 2014 , 2017 ), information retrieval services ( Wu et al, 2020a , 2021b ), human activity recognition ( Qiu et al, 2022 ), colorectal polyp region extraction ( Hu K. et al, 2022 ), location-based services ( Wu et al, 2020b , 2021a ), text clustering ( Guan et al, 2020 ), essay recommendation ( Liang et al, 2021 ), image denoising ( Zhang et al, 2020 ), drug-disease associations prediction ( Cai et al, 2021 ), other disease image segmentation ( Qi et al, 2022 ; Ren et al, 2022 ; Su et al, 2022 ), dynamic module detection ( Ma et al, 2020 ; Li D. et al, 2021 ), drug discovery ( Zhu F. et al, 2018 ; Li Y. et al, 2020 ), and road network planning ( Huang et al, 2022 ).…”

Section: Discussionmentioning

confidence: 99%

Multi-threshold image segmentation for melanoma based on Kapur’s entropy using enhanced ant colony optimization

Xiao

Ye²,

Zhao³

et al. 2022

Front. Neuroinform.

View full text Add to dashboard Cite

Melanoma is a malignant tumor formed by the cancerous transformation of melanocytes, and its medical images contain much information. However, the percentage of the critical information in the image is small, and the noise is non-uniformly distributed. We propose a new multi-threshold image segmentation model based on the two-dimensional histogram approach to the above problem. We present an enhanced ant colony optimization for continuous domains (EACOR) in the proposed model based on the soft besiege and chase strategies. Further, EACOR is combined with two-dimensional Kapur’s entropy to search for the optimal thresholds. An experiment on the IEEE CEC2014 benchmark function was conducted to measure the reliable global search capability of the EACOR algorithm in the proposed model. Moreover, we have also conducted several sets of experiments to test the validity of the image segmentation model proposed in this paper. The experimental results show that the segmented images from the proposed model outperform the comparison method in several evaluation metrics. Ultimately, the model proposed in this paper can provide high-quality samples for subsequent analysis of melanoma pathology images.

show abstract

Section: Discussionmentioning

confidence: 99%

Multi-threshold image segmentation for melanoma based on Kapur’s entropy using enhanced ant colony optimization

Xiao

Ye²,

Zhao³

et al. 2022

Front. Neuroinform.

View full text Add to dashboard Cite

show abstract

“…Semantic structures of objects and images are inherently low rank [27]. Recently, methods for low-rank matrix approximation have been developed to characterize the lowrank structures in images [7,12,28,33,29]. In this paper, we propose to formulate the problem of removing adversarial noise from attacked images while preserving important semantic structure information for successful recognition as a low-rank matrix completion problem.…”

Section: Low-rank Image Completion Based On Nuclear Norm Minimizationmentioning

confidence: 99%

Structure-Preserving Progressive Low-rank Image Completion for Defending Adversarial Attacks

Zhao¹,

Wang²,

Sun³

et al. 2021

Preprint

View full text Add to dashboard Cite

Deep neural networks recognize objects by analyzing local image details and summarizing their information along the inference layers to derive the final decision. Because of this, they are prone to adversarial attacks. Small sophisticated noise in the input images can accumulate along the network inference path and produce wrong decisions at the network output. On the other hand, human eyes recognize objects based on their global structure and semantic cues, instead of local image textures. Because of this, human eyes can still clearly recognize objects from images which have been heavily damaged by adversarial attacks. This leads to a very interesting approach for defending deep neural networks against adversarial attacks. In this work, we propose to develop a structure-preserving progressive low-rank image completion (SPLIC) method to remove unneeded texture details from the input images and shift the bias of deep neural networks towards global object structures and semantic cues. We formulate the problem into a low-rank matrix completion problem with progressively smoothed rank functions to avoid local minimums during the optimization process. Our experimental results demonstrate that the proposed method is able to successfully remove the insignificant local image details while preserving important global object structures. On black-box, gray-box, and whitebox attacks, our method outperforms existing defense methods (by up to 12.6%) and significantly improves the adversarial robustness of the network.

show abstract

“…Among them, JDDA [13] proposed two discriminative feature learning methods which jointly align domains and discriminative feature learning, resulting in better intra-class compactness and inter-class separability. Motivated by JDDA, we proposed an novel framework for cross-domain 3D objects retrieval by implementing the discriminative feature learning and feature transformation [28].…”

Section: B Deep Domain Adaptationmentioning

confidence: 99%

End-to-End Visual Domain Adaptation Network for Cross-Domain 3D CPS Data Retrieval

et al. 2019

View full text Add to dashboard Cite

3D CPS (Cyber Physical System) data has been widely generated and utilized for multiple applications, e.g. autonomous driving, unmanned aerial vehicle and so on. For large-scale 3D CPS data analysis, 3D object retrieval plays a significant role for urban perception. In this paper, we propose an endto-end domain adaptation framework for cross-domain 3D objects retrieval (C3DOR-Net), which learns a joint embedding space for 3D objects from different domains in an end-to-end manner. Specifically, we focus on the unsupervised case when 3D objects in the target domain are unlabeled. To better encode a 3D object, the proposed method learns multi-view visual features in a data-driven manner for 3D object representation. Then, the domain adaptation strategy is implemented to benefit both domain alignment and final classification. Specifically, an center-based discriminative feature learning method enables the domain invariant features with better intra-class compactness and inter-class separability. C3DOR-Net can achieve remarkable retrieval performances by maximizing the inter-class divergence and minimizing the intra-class divergence. We evaluate our method on two cross-domain protocols: 1) CAD-to-CAD object retrieval on two popular 3D datasets (NTU and PSB) in three designed cross-domain scenarios; 2) SHREC'19 monocular image based 3D object retrieval. Experimental results demonstrate that our method can significantly boost the cross-domain retrieval performances. INDEX TERMS 3D object retrieval, cross-domain learning, unsupervised domain adaptation.

show abstract

Exemplar-Based Denoising: A Unified Low-Rank Recovery Framework

Cited by 43 publications

References 26 publications

Multi-threshold image segmentation for melanoma based on Kapur’s entropy using enhanced ant colony optimization

Multi-threshold image segmentation for melanoma based on Kapur’s entropy using enhanced ant colony optimization

Structure-Preserving Progressive Low-rank Image Completion for Defending Adversarial Attacks

End-to-End Visual Domain Adaptation Network for Cross-Domain 3D CPS Data Retrieval

Contact Info

Product

Resources

About