Structure-aware depth super-resolution using Gaussian mixture model

Sunok Kim, Changjae Oh, Youngjung Kim, Kwanghoon Sohn

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

This paper presents a probabilistic optimization approach to enhance the resolution of a depth map. Conventionally, a high-resolution color image is considered as a cue for depth super-resolution under the assumption that the pixels with similar color likely belong to similar depth. This assumption might induce a texture transferring from the color image into the depth map and an edge blurring artifact to the depth boundaries. In order to alleviate these problems, we propose an efficient depth prior exploiting a Gaussian mixture model in which an estimated depth map is considered to a feature for computing affinity between two pixels. Furthermore, a fixed-point iteration scheme is adopted to address the non-linearity of a constraint derived from the proposed prior. The experimental results show that the proposed method outperforms state-of-the-art methods both quantitatively and qualitatively.

Original languageEnglish
Title of host publicationProceedings of SPIE-IS and T Electronic Imaging - Three-Dimensional Image Processing, Measurement (3DIPM), and Applications 2015
EditorsRobert Sitnik, William Puech
PublisherSPIE
ISBN (Electronic)9781628414837
DOIs
Publication statusPublished - 2015 Jan 1
EventThree-Dimensional Image Processing, Measurement (3DIPM), and Applications 2015 - San Francisco, United States
Duration: 2015 Feb 102015 Feb 12

Publication series

NameProceedings of SPIE - The International Society for Optical Engineering
Volume9393
ISSN (Print)0277-786X
ISSN (Electronic)1996-756X

Other

OtherThree-Dimensional Image Processing, Measurement (3DIPM), and Applications 2015
CountryUnited States
CitySan Francisco
Period15/2/1015/2/12

Fingerprint

Super-resolution
Gaussian Mixture Model
Depth Map
Color
Color Image
Pixels
Pixel
Fixed Point Iteration
Iteration Scheme
color
Textures
Affine transformation
Texture
pixels
High Resolution
Likely
Nonlinearity
blurring
cues
Optimization

All Science Journal Classification (ASJC) codes

  • Electronic, Optical and Magnetic Materials
  • Condensed Matter Physics
  • Computer Science Applications
  • Applied Mathematics
  • Electrical and Electronic Engineering

Cite this

Kim, S., Oh, C., Kim, Y., & Sohn, K. (2015). Structure-aware depth super-resolution using Gaussian mixture model. In R. Sitnik, & W. Puech (Eds.), Proceedings of SPIE-IS and T Electronic Imaging - Three-Dimensional Image Processing, Measurement (3DIPM), and Applications 2015 [93930J] (Proceedings of SPIE - The International Society for Optical Engineering; Vol. 9393). SPIE. https://doi.org/10.1117/12.2078795
Kim, Sunok ; Oh, Changjae ; Kim, Youngjung ; Sohn, Kwanghoon. / Structure-aware depth super-resolution using Gaussian mixture model. Proceedings of SPIE-IS and T Electronic Imaging - Three-Dimensional Image Processing, Measurement (3DIPM), and Applications 2015. editor / Robert Sitnik ; William Puech. SPIE, 2015. (Proceedings of SPIE - The International Society for Optical Engineering).
@inproceedings{b6795bd85403497e96318d7946d955d9,
title = "Structure-aware depth super-resolution using Gaussian mixture model",
abstract = "This paper presents a probabilistic optimization approach to enhance the resolution of a depth map. Conventionally, a high-resolution color image is considered as a cue for depth super-resolution under the assumption that the pixels with similar color likely belong to similar depth. This assumption might induce a texture transferring from the color image into the depth map and an edge blurring artifact to the depth boundaries. In order to alleviate these problems, we propose an efficient depth prior exploiting a Gaussian mixture model in which an estimated depth map is considered to a feature for computing affinity between two pixels. Furthermore, a fixed-point iteration scheme is adopted to address the non-linearity of a constraint derived from the proposed prior. The experimental results show that the proposed method outperforms state-of-the-art methods both quantitatively and qualitatively.",
author = "Sunok Kim and Changjae Oh and Youngjung Kim and Kwanghoon Sohn",
year = "2015",
month = "1",
day = "1",
doi = "10.1117/12.2078795",
language = "English",
series = "Proceedings of SPIE - The International Society for Optical Engineering",
publisher = "SPIE",
editor = "Robert Sitnik and William Puech",
booktitle = "Proceedings of SPIE-IS and T Electronic Imaging - Three-Dimensional Image Processing, Measurement (3DIPM), and Applications 2015",
address = "United States",

}

Kim, S, Oh, C, Kim, Y & Sohn, K 2015, Structure-aware depth super-resolution using Gaussian mixture model. in R Sitnik & W Puech (eds), Proceedings of SPIE-IS and T Electronic Imaging - Three-Dimensional Image Processing, Measurement (3DIPM), and Applications 2015., 93930J, Proceedings of SPIE - The International Society for Optical Engineering, vol. 9393, SPIE, Three-Dimensional Image Processing, Measurement (3DIPM), and Applications 2015, San Francisco, United States, 15/2/10. https://doi.org/10.1117/12.2078795

Structure-aware depth super-resolution using Gaussian mixture model. / Kim, Sunok; Oh, Changjae; Kim, Youngjung; Sohn, Kwanghoon.

Proceedings of SPIE-IS and T Electronic Imaging - Three-Dimensional Image Processing, Measurement (3DIPM), and Applications 2015. ed. / Robert Sitnik; William Puech. SPIE, 2015. 93930J (Proceedings of SPIE - The International Society for Optical Engineering; Vol. 9393).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Structure-aware depth super-resolution using Gaussian mixture model

AU - Kim, Sunok

AU - Oh, Changjae

AU - Kim, Youngjung

AU - Sohn, Kwanghoon

PY - 2015/1/1

Y1 - 2015/1/1

N2 - This paper presents a probabilistic optimization approach to enhance the resolution of a depth map. Conventionally, a high-resolution color image is considered as a cue for depth super-resolution under the assumption that the pixels with similar color likely belong to similar depth. This assumption might induce a texture transferring from the color image into the depth map and an edge blurring artifact to the depth boundaries. In order to alleviate these problems, we propose an efficient depth prior exploiting a Gaussian mixture model in which an estimated depth map is considered to a feature for computing affinity between two pixels. Furthermore, a fixed-point iteration scheme is adopted to address the non-linearity of a constraint derived from the proposed prior. The experimental results show that the proposed method outperforms state-of-the-art methods both quantitatively and qualitatively.

AB - This paper presents a probabilistic optimization approach to enhance the resolution of a depth map. Conventionally, a high-resolution color image is considered as a cue for depth super-resolution under the assumption that the pixels with similar color likely belong to similar depth. This assumption might induce a texture transferring from the color image into the depth map and an edge blurring artifact to the depth boundaries. In order to alleviate these problems, we propose an efficient depth prior exploiting a Gaussian mixture model in which an estimated depth map is considered to a feature for computing affinity between two pixels. Furthermore, a fixed-point iteration scheme is adopted to address the non-linearity of a constraint derived from the proposed prior. The experimental results show that the proposed method outperforms state-of-the-art methods both quantitatively and qualitatively.

UR - http://www.scopus.com/inward/record.url?scp=84928485306&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84928485306&partnerID=8YFLogxK

U2 - 10.1117/12.2078795

DO - 10.1117/12.2078795

M3 - Conference contribution

AN - SCOPUS:84928485306

T3 - Proceedings of SPIE - The International Society for Optical Engineering

BT - Proceedings of SPIE-IS and T Electronic Imaging - Three-Dimensional Image Processing, Measurement (3DIPM), and Applications 2015

A2 - Sitnik, Robert

A2 - Puech, William

PB - SPIE

ER -

Kim S, Oh C, Kim Y, Sohn K. Structure-aware depth super-resolution using Gaussian mixture model. In Sitnik R, Puech W, editors, Proceedings of SPIE-IS and T Electronic Imaging - Three-Dimensional Image Processing, Measurement (3DIPM), and Applications 2015. SPIE. 2015. 93930J. (Proceedings of SPIE - The International Society for Optical Engineering). https://doi.org/10.1117/12.2078795