Single Image Layer Separation via Deep Admm Unrolling

Author(s):  
Risheng Liu ◽  
Zhiying Jiang ◽  
Xin Fan ◽  
Haojie Li ◽  
Zhongxuan Luo
2020 ◽  
Vol 34 (07) ◽  
pp. 11661-11668 ◽  
Author(s):  
Yunfei Liu ◽  
Feng Lu

Many real world vision tasks, such as reflection removal from a transparent surface and intrinsic image decomposition, can be modeled as single image layer separation. However, this problem is highly ill-posed, requiring accurately aligned and hard to collect triplet data to train the CNN models. To address this problem, this paper proposes an unsupervised method that requires no ground truth data triplet in training. At the core of the method are two assumptions about data distributions in the latent spaces of different layers, based on which a novel unsupervised layer separation pipeline can be derived. Then the method can be constructed based on the GANs framework with self-supervision and cycle consistency constraints, etc. Experimental results demonstrate its successfulness in outperforming existing unsupervised methods in both synthetic and real world tasks. The method also shows its ability to solve a more challenging multi-layer separation task.


IEEE Access ◽  
2019 ◽  
Vol 7 ◽  
pp. 178685-178698 ◽  
Author(s):  
Chenggang Dai ◽  
Mingxing Lin ◽  
Jingkun Wang ◽  
Xiao Hu

2020 ◽  
Vol 31 (5) ◽  
pp. 1653-1666 ◽  
Author(s):  
Risheng Liu ◽  
Zhiying Jiang ◽  
Xin Fan ◽  
Zhongxuan Luo

2014 ◽  
Vol 20 (10) ◽  
pp. 1837-1841
Author(s):  
Nurulfajar Abd Manap ◽  
Masrullizam Mat Ibrahim ◽  
John Soraghan ◽  
Lykourgos Petropoulakis

VASA ◽  
2015 ◽  
Vol 44 (2) ◽  
pp. 122-128 ◽  
Author(s):  
Mandy Becker ◽  
Tom Schilling ◽  
Olga von Beckerath ◽  
Knut Kröger

Background: To clarify the clinical use of sonography for differentiation of edema we tried to answer the question whether a group of doctors can differentiate lymphedema from cardiac, hepatic or venous edema just by analysing sonographic images of the edema. Patients and methods: 38 (70 ± 12 years, 22 (58 %) females) patients with lower limb edema were recruited according the clinical diagnosis: 10 (26 %) lymphedema, 16 (42 %) heart insufficiency, 6 (16 %) venous disorders, 6 (16 %) chronic hepatic disease. Edema was depicted sonographically at the most affected leg in a standardised way at distal and proximal calf. 38 sets of images were anonymised and send to 5 experienced doctors. They were asked whether they can see criteria for lymphedema: 1. anechoic gaps, 2. horizontal gaps and 3. echoic rims. Results: Accepting an edema as lymphedema if only one doctor sees at least one of the three criteria for lymphatic edema on each single image all edema would be classified as lymphatic. Accepting lymphedema only if all doctors see at least one of the three criteria on the distal image of the same patient 80 % of the patients supposed to have lymphedema are classified as such, but also the majority of cardiac, venous and hepatic edema. Accepting lymphedema only if all doctors see all three criteria on the distal image of the same patients no edema would be classified as lymphatic. In addition we separated patients by Stemmers’ sign in those with positive and negative sign. The interpretation of the images was not different between both groups. Conclusions: Our analysis shows that it is not possible to differentiate lymphedema from other lower limb edema sonographically.


Sign in / Sign up

Export Citation Format

Share Document