Single image reflection removal via learning with multi-image constraints

Reflections are very common phenomena in our daily photography, which distract people's attention from the scene behind the glass. The problem of removing reflection artifacts is important but challenging due to its ill-posed nature. The traditional approaches solve an optimization problem over...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yin, Yingda, Fan, Qingnan, Chen, Dongdong, Wang, Yujie, Aviles-Rivero, Angelica, Li, Ruoteng, Schnlieb, Carola-Bibiane, Chen, Baoquan
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Yin, Yingda
Fan, Qingnan
Chen, Dongdong
Wang, Yujie
Aviles-Rivero, Angelica
Li, Ruoteng
Schnlieb, Carola-Bibiane
Chen, Baoquan
description Reflections are very common phenomena in our daily photography, which distract people's attention from the scene behind the glass. The problem of removing reflection artifacts is important but challenging due to its ill-posed nature. The traditional approaches solve an optimization problem over the constraints induced from multiple images, at the expense of large computation costs. Recent learning-based approaches have demonstrated a significant improvement in both performance and running time for single image reflection removal, but are limited as they require a large number of synthetic reflection/clean image pairs for direct supervision to approximate the ground truth, at the risk of overfitting in the synthetic image domain and degrading in the real image domain. In this paper, we propose a novel learning-based solution that combines the advantages of the aforementioned approaches and overcomes their drawbacks. Our algorithm works by learning a deep neural network to optimize the target with joint constraints enhanced among multiple input images during the training phase, but is able to eliminate reflections only from a single input for evaluation. Our algorithm runs in real-time and achieves state-of-the-art reflection removal performance on real images. We further propose a strong network backbone that disentangles the background and reflection information into separate latent codes, which are embedded into a shared one-branch deep neural network for both background and reflection predictions. The proposed backbone experimentally performs better than the other common network implementations, and provides insightful knowledge to understand the reflection removal task.
doi_str_mv 10.48550/arxiv.1912.03623
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_1912_03623</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>1912_03623</sourcerecordid><originalsourceid>FETCH-LOGICAL-a673-6094f0235c2386edf8d255a22b901979d85e0fc7e087806f0d0af918c2ad5bb63</originalsourceid><addsrcrecordid>eNotj8tOwzAURL1hgVo-gBX-gYQbu3bsZVXxqFSJRbuPbvwolhwHOSbA3xNaVjOLmdEcQu4bqDdKCHjE_B3mutENq4FLxm_J_hjSOToaBjw7mp2PzpQwpsUO44yRzgFpdJjTkqNfobzT4TOWUF0LZkxTyRhSmdbkxmOc3N2_rsjp-em0e60Oby_73fZQoWx5JUFvPDAuDONKOuuVZUIgY72GRrfaKuHAm9aBahVIDxbQ60YZhlb0veQr8nCdvbB0H3k5kn-6P6buwsR_ATPdR1Y</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Single image reflection removal via learning with multi-image constraints</title><source>arXiv.org</source><creator>Yin, Yingda ; Fan, Qingnan ; Chen, Dongdong ; Wang, Yujie ; Aviles-Rivero, Angelica ; Li, Ruoteng ; Schnlieb, Carola-Bibiane ; Chen, Baoquan</creator><creatorcontrib>Yin, Yingda ; Fan, Qingnan ; Chen, Dongdong ; Wang, Yujie ; Aviles-Rivero, Angelica ; Li, Ruoteng ; Schnlieb, Carola-Bibiane ; Chen, Baoquan</creatorcontrib><description>Reflections are very common phenomena in our daily photography, which distract people's attention from the scene behind the glass. The problem of removing reflection artifacts is important but challenging due to its ill-posed nature. The traditional approaches solve an optimization problem over the constraints induced from multiple images, at the expense of large computation costs. Recent learning-based approaches have demonstrated a significant improvement in both performance and running time for single image reflection removal, but are limited as they require a large number of synthetic reflection/clean image pairs for direct supervision to approximate the ground truth, at the risk of overfitting in the synthetic image domain and degrading in the real image domain. In this paper, we propose a novel learning-based solution that combines the advantages of the aforementioned approaches and overcomes their drawbacks. Our algorithm works by learning a deep neural network to optimize the target with joint constraints enhanced among multiple input images during the training phase, but is able to eliminate reflections only from a single input for evaluation. Our algorithm runs in real-time and achieves state-of-the-art reflection removal performance on real images. We further propose a strong network backbone that disentangles the background and reflection information into separate latent codes, which are embedded into a shared one-branch deep neural network for both background and reflection predictions. The proposed backbone experimentally performs better than the other common network implementations, and provides insightful knowledge to understand the reflection removal task.</description><identifier>DOI: 10.48550/arxiv.1912.03623</identifier><language>eng</language><subject>Computer Science - Computer Vision and Pattern Recognition</subject><creationdate>2019-12</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/1912.03623$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.1912.03623$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Yin, Yingda</creatorcontrib><creatorcontrib>Fan, Qingnan</creatorcontrib><creatorcontrib>Chen, Dongdong</creatorcontrib><creatorcontrib>Wang, Yujie</creatorcontrib><creatorcontrib>Aviles-Rivero, Angelica</creatorcontrib><creatorcontrib>Li, Ruoteng</creatorcontrib><creatorcontrib>Schnlieb, Carola-Bibiane</creatorcontrib><creatorcontrib>Chen, Baoquan</creatorcontrib><title>Single image reflection removal via learning with multi-image constraints</title><description>Reflections are very common phenomena in our daily photography, which distract people's attention from the scene behind the glass. The problem of removing reflection artifacts is important but challenging due to its ill-posed nature. The traditional approaches solve an optimization problem over the constraints induced from multiple images, at the expense of large computation costs. Recent learning-based approaches have demonstrated a significant improvement in both performance and running time for single image reflection removal, but are limited as they require a large number of synthetic reflection/clean image pairs for direct supervision to approximate the ground truth, at the risk of overfitting in the synthetic image domain and degrading in the real image domain. In this paper, we propose a novel learning-based solution that combines the advantages of the aforementioned approaches and overcomes their drawbacks. Our algorithm works by learning a deep neural network to optimize the target with joint constraints enhanced among multiple input images during the training phase, but is able to eliminate reflections only from a single input for evaluation. Our algorithm runs in real-time and achieves state-of-the-art reflection removal performance on real images. We further propose a strong network backbone that disentangles the background and reflection information into separate latent codes, which are embedded into a shared one-branch deep neural network for both background and reflection predictions. The proposed backbone experimentally performs better than the other common network implementations, and provides insightful knowledge to understand the reflection removal task.</description><subject>Computer Science - Computer Vision and Pattern Recognition</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotj8tOwzAURL1hgVo-gBX-gYQbu3bsZVXxqFSJRbuPbvwolhwHOSbA3xNaVjOLmdEcQu4bqDdKCHjE_B3mutENq4FLxm_J_hjSOToaBjw7mp2PzpQwpsUO44yRzgFpdJjTkqNfobzT4TOWUF0LZkxTyRhSmdbkxmOc3N2_rsjp-em0e60Oby_73fZQoWx5JUFvPDAuDONKOuuVZUIgY72GRrfaKuHAm9aBahVIDxbQ60YZhlb0veQr8nCdvbB0H3k5kn-6P6buwsR_ATPdR1Y</recordid><startdate>20191208</startdate><enddate>20191208</enddate><creator>Yin, Yingda</creator><creator>Fan, Qingnan</creator><creator>Chen, Dongdong</creator><creator>Wang, Yujie</creator><creator>Aviles-Rivero, Angelica</creator><creator>Li, Ruoteng</creator><creator>Schnlieb, Carola-Bibiane</creator><creator>Chen, Baoquan</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20191208</creationdate><title>Single image reflection removal via learning with multi-image constraints</title><author>Yin, Yingda ; Fan, Qingnan ; Chen, Dongdong ; Wang, Yujie ; Aviles-Rivero, Angelica ; Li, Ruoteng ; Schnlieb, Carola-Bibiane ; Chen, Baoquan</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a673-6094f0235c2386edf8d255a22b901979d85e0fc7e087806f0d0af918c2ad5bb63</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Computer Science - Computer Vision and Pattern Recognition</topic><toplevel>online_resources</toplevel><creatorcontrib>Yin, Yingda</creatorcontrib><creatorcontrib>Fan, Qingnan</creatorcontrib><creatorcontrib>Chen, Dongdong</creatorcontrib><creatorcontrib>Wang, Yujie</creatorcontrib><creatorcontrib>Aviles-Rivero, Angelica</creatorcontrib><creatorcontrib>Li, Ruoteng</creatorcontrib><creatorcontrib>Schnlieb, Carola-Bibiane</creatorcontrib><creatorcontrib>Chen, Baoquan</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Yin, Yingda</au><au>Fan, Qingnan</au><au>Chen, Dongdong</au><au>Wang, Yujie</au><au>Aviles-Rivero, Angelica</au><au>Li, Ruoteng</au><au>Schnlieb, Carola-Bibiane</au><au>Chen, Baoquan</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Single image reflection removal via learning with multi-image constraints</atitle><date>2019-12-08</date><risdate>2019</risdate><abstract>Reflections are very common phenomena in our daily photography, which distract people's attention from the scene behind the glass. The problem of removing reflection artifacts is important but challenging due to its ill-posed nature. The traditional approaches solve an optimization problem over the constraints induced from multiple images, at the expense of large computation costs. Recent learning-based approaches have demonstrated a significant improvement in both performance and running time for single image reflection removal, but are limited as they require a large number of synthetic reflection/clean image pairs for direct supervision to approximate the ground truth, at the risk of overfitting in the synthetic image domain and degrading in the real image domain. In this paper, we propose a novel learning-based solution that combines the advantages of the aforementioned approaches and overcomes their drawbacks. Our algorithm works by learning a deep neural network to optimize the target with joint constraints enhanced among multiple input images during the training phase, but is able to eliminate reflections only from a single input for evaluation. Our algorithm runs in real-time and achieves state-of-the-art reflection removal performance on real images. We further propose a strong network backbone that disentangles the background and reflection information into separate latent codes, which are embedded into a shared one-branch deep neural network for both background and reflection predictions. The proposed backbone experimentally performs better than the other common network implementations, and provides insightful knowledge to understand the reflection removal task.</abstract><doi>10.48550/arxiv.1912.03623</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.1912.03623
ispartof
issn
language eng
recordid cdi_arxiv_primary_1912_03623
source arXiv.org
subjects Computer Science - Computer Vision and Pattern Recognition
title Single image reflection removal via learning with multi-image constraints
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-02T00%3A17%3A37IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Single%20image%20reflection%20removal%20via%20learning%20with%20multi-image%20constraints&rft.au=Yin,%20Yingda&rft.date=2019-12-08&rft_id=info:doi/10.48550/arxiv.1912.03623&rft_dat=%3Carxiv_GOX%3E1912_03623%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true