Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Comprehensive Overview of Fish-Eye Camera Distortion Correction Methods

Jian Xu, De-Wei Han, Kang Li, Jun-Jie Li, Zhao-Yuan Ma

TL;DR

The paper surveys a wide range of fisheye distortion correction methods, ranging from classical polynomial models such as the Brown-Conrady and Kannala-Brandt frameworks to direct, panorama-based, and deep learning approaches. It highlights projection models (e.g., Equidistant, Equiangular, Orthographic, and Stereographic) that underpin distortion generation and correction, and it categorizes methods into feature-based, direct, and data-driven techniques, including CNNs, GANs, and ViTs. A key contribution is the synthesis of model-driven calibration with recent deep learning advances, discussing their relative advantages, limitations, and practical considerations such as dataset availability and evaluation metrics like PSNR and SSIM. The work emphasizes that no universal dataset exists, urging careful dataset design and metric selection to enable robust comparison and real-world deployment of fisheye correction methods.

Abstract

The fisheye camera, with its unique wide field of view and other characteristics, has found extensive applications in various fields. However, the fisheye camera suffers from significant distortion compared to pinhole cameras, resulting in distorted images of captured objects. Fish-eye camera distortion is a common issue in digital image processing, requiring effective correction techniques to enhance image quality. This review provides a comprehensive overview of various methods used for fish-eye camera distortion correction. The article explores the polynomial distortion model, which utilizes polynomial functions to model and correct radial distortions. Additionally, alternative approaches such as panorama mapping, grid mapping, direct methods, and deep learning-based methods are discussed. The review highlights the advantages, limitations, and recent advancements of each method, enabling readers to make informed decisions based on their specific needs.

A Comprehensive Overview of Fish-Eye Camera Distortion Correction Methods

TL;DR

The paper surveys a wide range of fisheye distortion correction methods, ranging from classical polynomial models such as the Brown-Conrady and Kannala-Brandt frameworks to direct, panorama-based, and deep learning approaches. It highlights projection models (e.g., Equidistant, Equiangular, Orthographic, and Stereographic) that underpin distortion generation and correction, and it categorizes methods into feature-based, direct, and data-driven techniques, including CNNs, GANs, and ViTs. A key contribution is the synthesis of model-driven calibration with recent deep learning advances, discussing their relative advantages, limitations, and practical considerations such as dataset availability and evaluation metrics like PSNR and SSIM. The work emphasizes that no universal dataset exists, urging careful dataset design and metric selection to enable robust comparison and real-world deployment of fisheye correction methods.

Abstract

The fisheye camera, with its unique wide field of view and other characteristics, has found extensive applications in various fields. However, the fisheye camera suffers from significant distortion compared to pinhole cameras, resulting in distorted images of captured objects. Fish-eye camera distortion is a common issue in digital image processing, requiring effective correction techniques to enhance image quality. This review provides a comprehensive overview of various methods used for fish-eye camera distortion correction. The article explores the polynomial distortion model, which utilizes polynomial functions to model and correct radial distortions. Additionally, alternative approaches such as panorama mapping, grid mapping, direct methods, and deep learning-based methods are discussed. The review highlights the advantages, limitations, and recent advancements of each method, enabling readers to make informed decisions based on their specific needs.
Paper Structure (30 sections, 22 equations, 2 figures)

This paper contains 30 sections, 22 equations, 2 figures.

Table of Contents

  1. Introduction
  2. Camera Projection Models
  3. Equidistant Projection Model
  4. Equiangular Projection Model
  5. Orthographic Projection Model
  6. Stereographic Projection Model
  7. Distortion Correction Methods
  8. Distortion Types
  9. Radial Distortions
  10. Tangential (De-centering) Distortions
  11. Polynomial Distortion Model
  12. Brown-Conrady
  13. Kannala-Brandt
  14. Feature-based Methods
  15. Corner Detection and Rectification
  16. ...and 15 more sections

Figures (2)

  • Figure 1: (a) Projections with f=1 ((i) to (v) are representing the five projection models mentioned above). (b) Fish-eye camera model.kannala_GenericCameraCalibration_2004
  • Figure 2: Distribution of the camera parameters for dataset.wakai2021rethinking