Mid Sweden University

miun.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Area-Based Depth Estimation for Monochromatic Feature-Sparse Orthographic Capture
Mid Sweden University, Faculty of Science, Technology and Media, Department of Information Systems and Technology. (Realistic3D)
Department of Optics, University of Valencia, Burjassot, Spain.
Mid Sweden University, Faculty of Science, Technology and Media, Department of Information Systems and Technology.
Department of Optics, University of Valencia, Burjassot, Spain.
2018 (English)In: 2018 26th European Signal Processing Conference (EUSIPCO), IEEE conference proceedings, 2018, p. 206-210, article id 8553336Conference paper, Published paper (Refereed)
Abstract [en]

With the rapid development of light field technology, depth estimation has been highlighted as one of the critical problems in the field, and a number of approaches have been proposed to extract the depth of the scene. However, depthestimation by stereo matching becomes difficult and unreliable when the captured images lack both color and feature information. In this paper, we propose a scheme that extracts robust depth from monochromatic, feature-sparse scenes recorded in orthographic sub-aperture images. Unlike approaches which relyon the rich color and texture information across the sub-aperture views, our approach is based on depth from focus techniques. First, we superimpose shifted sub-aperture images on top of anarbitrarily chosen central image. To focus on different depths, the shift amount is varied based on the micro-lens array properties. Next, an area-based depth estimation approach is applied tofind the best match among the focal stack and generate the dense depth map. This process is repeated for each sub-aperture image. Finally, occlusions are handled by merging depth maps generated from different central images followed by a voting process. Results show that the proposed scheme is more suitable than conventional depth estimation approaches in the context of orthographic captures that have insufficient color and feature information, such as microscopic fluorescence imaging.

Place, publisher, year, edition, pages
IEEE conference proceedings, 2018. p. 206-210, article id 8553336
Keywords [en]
Depth estimation, integral imaging, orthographic views, depth from focus
National Category
Computer Sciences
Identifiers
URN: urn:nbn:se:miun:diva-34418ISI: 000455614900042Scopus ID: 2-s2.0-85059811493OAI: oai:DiVA.org:miun-34418DiVA, id: diva2:1248223
Conference
EUSIPCO 2018, 26th European Signal Processing Conference, Rome, Italy, September 3-7, 2018
Available from: 2018-09-14 Created: 2018-09-14 Last updated: 2020-05-11Bibliographically approved
In thesis
1. Computational Light Field Photography: Depth Estimation, Demosaicing, and Super-Resolution
Open this publication in new window or tab >>Computational Light Field Photography: Depth Estimation, Demosaicing, and Super-Resolution
2020 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]

The transition of camera technology from film-based cameras to digital cameras has been witnessed in the past twenty years, along with impressive technological advances in processing massively digitized media content. Today, a new evolution emerged -- the migration from 2D content to immersive perception. This rising trend has a profound and long-term impact to our society, fostering technologies such as teleconferencing and remote surgery. The trend is also reflected in the scientific research community, and more intention has been drawn to the light field and its applications.

 

The purpose of this dissertation is to develop a better understanding of light field structure by analyzing its sampling behavior and to addresses three problems concerning the light field processing pipeline: 1) How to address the depth estimation problem when there is limited color and texture information. 2) How to improve the rendered image quality by using the inherent depth information. 3) How to solve the interdependence conflict of demosaicing and depth estimation.

 

The first problem is solved by a hybrid depth estimation approach that combines advantages of correspondence matching and depth-from-focus, where occlusion is handled by involving multiple depth maps in a voting scheme. The second problem is divided into two specific tasks -- demosaicing and super-resolution, where depth-assisted light field analysis is employed to surpass the competence of traditional image processing. The third problem is tackled with an inferential graph model that encodes the connections between demosaicing and depth estimation explicitly, and jointly performs a global optimization for both tasks.

 

The proposed depth estimation approach shows a noticeable improvement in point clouds and depth maps, compared with references methods. Furthermore, the objective metrics and visual quality are compared with classical sensor-based demosaicing and multi-image super-resolution to show the effectiveness of the proposed depth-assisted light field processing methods. Finally, a multi-task graph model is proposed to challenge the performance of the sequential light field image processing pipeline. The proposed method is validated with various kinds of light fields, and outperforms the state-of-the-art in both demosaicing and depth estimation tasks.

 

The works presented in this dissertation raise a novel view of the light field data structure in general, and provide tools to solve image processing problems in specific. The impact of the outcome can be manifold: To support scientific research with light field microscopes, to stabilize the performance of range cameras for industrial applications, as well as to provide individuals with a high-quality immersive experience.

Abstract [sv]

Under de senaste tjugo åren har det skett en övergång från filmbaserad till digital kamerateknik, parallellt med en imponerande teknisk utveckling inom bearbetning av omfattande digitaliserat medieinnehåll. På senare tid även en ny utvecklingslinje – övergången från 2D-innehåll till omslutande perception. Detta är en utveckling som har långtgående och långvarig påverkan på samhället och främjar arbetsmetoder såsom telekonferens och fjärrstyrd kirurgi. Den här utvecklingst trenden återspeglas också i det vetenskapliga forskningssamhället, och mer uppmärksamhet har lagts på light field och dess olika tillämpningsområden.

Syftet med avhandlingen är att nå en bättre förståelse av strukturen i light field genom att analysera hur light field samplas, och att lösa tre problem inom behandlingsprocessen av light field: 1) Hur problemet med djupestimering kan lösas med begränsad information om färg och textur. 2) Hur renderad bildkvalitet kan förbättras genom att utnyttja den inneboende djupinformationen. 3) Hur beroendekonflikten mellan demosaicing (färgfiltrering) och djupestimering kan lösas.

Det första problemet har lösts genom en hybridmetod för djupestimering, som kombinerar styrkorna med korrespondensmatchning och djup från fokus, där ocklusion hanteras genom att använda flera djupkartor i ett röstningssystem. Det andra problemet delas upp i två separata moment – demosaicing och superupplösning, där djupassisterad analys av light field används för att överträffa kapaciteten för traditionell bildbehandling. Det tredje problemet har angripits med en inferentiell grafmodell som explicit kopplar samman demosaicing och djupestimering, och samfällt utför en global optimering för båda dessa processteg.

Den metod för djupestimering som föreslås producerar visuellt tilltalande punktmoln och djupkartor, jämfört med andra referensmetoder. Objektiva mätvärden och visuell kvalitet jämförs vidare med klassisk sensorbaserad demosaicing och superupplösning från multipla bilder, för att visa effektiviteten hos de föreslagna metoderna för djupassisterad behandling av light field. En multitaskande grafmodell föreslås även för att matcha och överträffa prestandan hos sekventiell light field-baserad bildbehandling. Den metod som föreslås valideras med olika sorters light fields och överträffar de bästa existerande metoderna inom både demosaicing och djupestimering.

De arbeten som presenteras i avhandlingen utgör ett nytt sätt att betrakta den generella datastrukturen hos light field, och tillhandahåller verktyg för att lösa specifika bildbehandlingsproblem. Effekterna av dessa resultat kan vara många, till exempel som stöd för vetenskaplig forskning om light field-baserade mikroskop, för att förbättra prestandan hos avståndsmätande kameror i industriella tillämpningar, såväl som för att erbjuda högkvalitativa omslutande mediaupplevelser.

Place, publisher, year, edition, pages
Sundsvall: Mid Sweden University, 2020. p. 56
Series
Mid Sweden University doctoral thesis, ISSN 1652-893X ; 327
Keywords
Light field, computational photography, depth estimation, demosaicing, super-resolution
National Category
Computer and Information Sciences
Identifiers
urn:nbn:se:miun:diva-39005 (URN)978-91-88947-57-4 (ISBN)
Public defence
2020-06-10, C312, Holmgatan 10, Sundsvall, 09:00 (English)
Opponent
Supervisors
Projects
European Unions Horizon 2020 under the Marie Sklodowska-Curie grant agreement No 676401, European Training Network on Full Parallax Imaging
Note

Vid tidpunkten för disputationen var följande delarbeten opublicerade: delarbete 4 manuskript och delarbete 5 inskickat.

At the time of the doctoral defence the following papers were unpublished: paper 4 manuscript and paper 5 submitted.

Available from: 2020-05-18 Created: 2020-05-11 Last updated: 2020-05-13Bibliographically approved

Open Access in DiVA

fulltext(3390 kB)933 downloads
File information
File name FULLTEXT01.pdfFile size 3390 kBChecksum SHA-512
d38aae8e078ec91948063a7800593ea7c85d583efa7a111aed130f7e748addca7cdaa97760d09279f649d4184c6385b8606b15108511d469e8c5548ddddc7bce
Type fulltextMimetype application/pdf

Scopus

Authority records

Li, YongweiSjöström, Mårten

Search in DiVA

By author/editor
Li, YongweiSjöström, Mårten
By organisation
Department of Information Systems and Technology
Computer Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 933 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 1208 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf