Cross-Calibration of RGB and Thermal Cameras with a LIDAR for RGB-Depth-Thermal Mapping

Aravindhan K. Krishnan, Srikanth Saripalli

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

We present a method for calibrating the extrinsic parameters between a RGB camera, a thermal camera, and a LIDAR. The calibration procedure we use is common to both the RGB and thermal cameras. The extrinsic calibration procedure assumes that the cameras are geometrically calibrated. To aid the geometric calibration of the thermal camera, we use a calibration target made of black-and-white melamine that looks like a checkerboard pattern in the thermal and RGB images. For the extrinsic calibration, we place a circular calibration target in the common field of view of the cameras and the LIDAR and compute the extrinsic parameters by minimizing an objective function that aligns the edges of the circular target in the LIDAR to its corresponding edges in the RGB and thermal images. We illustrate the convexity of the objective function and discuss the convergence of the algorithm. We then identify the various sources of coloring errors (after cross-calibration) as (a) noise in the LIDAR points, (b) error in the intrinsic parameters of the camera, (c) error in the translation parameters between the LIDAR and the camera and (d) error in the rotation parameters between the LIDAR and the camera. We analyze the contribution of these errors with respect to the coloring of a 3D point. We illustrate that these errors are related to the depth of the 3D point considered - with errors (a), (b), and (c) being inversely proportional to the depth, and error (d) being directly proportional to the depth.

Original languageEnglish (US)
Pages (from-to)59-78
Number of pages20
JournalUnmanned Systems
Volume5
Issue number2
DOIs
StatePublished - Apr 1 2017
Externally publishedYes

Fingerprint

Calibration
Camera
Cameras
Coloring
Target
Colouring
Objective function
Directly proportional
Hot Temperature
Melamine
Field of View
Convexity

Keywords

  • Cross-calibration
  • RGB thermal mapping
  • thermal cameras

ASJC Scopus subject areas

  • Aerospace Engineering
  • Automotive Engineering
  • Control and Systems Engineering
  • Control and Optimization

Cite this

Cross-Calibration of RGB and Thermal Cameras with a LIDAR for RGB-Depth-Thermal Mapping. / Krishnan, Aravindhan K.; Saripalli, Srikanth.

In: Unmanned Systems, Vol. 5, No. 2, 01.04.2017, p. 59-78.

Research output: Contribution to journalArticle

Krishnan, Aravindhan K. ; Saripalli, Srikanth. / Cross-Calibration of RGB and Thermal Cameras with a LIDAR for RGB-Depth-Thermal Mapping. In: Unmanned Systems. 2017 ; Vol. 5, No. 2. pp. 59-78.
@article{e92343bf347848ceb9bb861fbf96ebc9,
title = "Cross-Calibration of RGB and Thermal Cameras with a LIDAR for RGB-Depth-Thermal Mapping",
abstract = "We present a method for calibrating the extrinsic parameters between a RGB camera, a thermal camera, and a LIDAR. The calibration procedure we use is common to both the RGB and thermal cameras. The extrinsic calibration procedure assumes that the cameras are geometrically calibrated. To aid the geometric calibration of the thermal camera, we use a calibration target made of black-and-white melamine that looks like a checkerboard pattern in the thermal and RGB images. For the extrinsic calibration, we place a circular calibration target in the common field of view of the cameras and the LIDAR and compute the extrinsic parameters by minimizing an objective function that aligns the edges of the circular target in the LIDAR to its corresponding edges in the RGB and thermal images. We illustrate the convexity of the objective function and discuss the convergence of the algorithm. We then identify the various sources of coloring errors (after cross-calibration) as (a) noise in the LIDAR points, (b) error in the intrinsic parameters of the camera, (c) error in the translation parameters between the LIDAR and the camera and (d) error in the rotation parameters between the LIDAR and the camera. We analyze the contribution of these errors with respect to the coloring of a 3D point. We illustrate that these errors are related to the depth of the 3D point considered - with errors (a), (b), and (c) being inversely proportional to the depth, and error (d) being directly proportional to the depth.",
keywords = "Cross-calibration, RGB thermal mapping, thermal cameras",
author = "Krishnan, {Aravindhan K.} and Srikanth Saripalli",
year = "2017",
month = "4",
day = "1",
doi = "10.1142/S2301385017500054",
language = "English (US)",
volume = "5",
pages = "59--78",
journal = "Unmanned Systems",
issn = "2301-3850",
publisher = "World Scientific Publishing Co. Pte Ltd",
number = "2",

}

TY - JOUR

T1 - Cross-Calibration of RGB and Thermal Cameras with a LIDAR for RGB-Depth-Thermal Mapping

AU - Krishnan, Aravindhan K.

AU - Saripalli, Srikanth

PY - 2017/4/1

Y1 - 2017/4/1

N2 - We present a method for calibrating the extrinsic parameters between a RGB camera, a thermal camera, and a LIDAR. The calibration procedure we use is common to both the RGB and thermal cameras. The extrinsic calibration procedure assumes that the cameras are geometrically calibrated. To aid the geometric calibration of the thermal camera, we use a calibration target made of black-and-white melamine that looks like a checkerboard pattern in the thermal and RGB images. For the extrinsic calibration, we place a circular calibration target in the common field of view of the cameras and the LIDAR and compute the extrinsic parameters by minimizing an objective function that aligns the edges of the circular target in the LIDAR to its corresponding edges in the RGB and thermal images. We illustrate the convexity of the objective function and discuss the convergence of the algorithm. We then identify the various sources of coloring errors (after cross-calibration) as (a) noise in the LIDAR points, (b) error in the intrinsic parameters of the camera, (c) error in the translation parameters between the LIDAR and the camera and (d) error in the rotation parameters between the LIDAR and the camera. We analyze the contribution of these errors with respect to the coloring of a 3D point. We illustrate that these errors are related to the depth of the 3D point considered - with errors (a), (b), and (c) being inversely proportional to the depth, and error (d) being directly proportional to the depth.

AB - We present a method for calibrating the extrinsic parameters between a RGB camera, a thermal camera, and a LIDAR. The calibration procedure we use is common to both the RGB and thermal cameras. The extrinsic calibration procedure assumes that the cameras are geometrically calibrated. To aid the geometric calibration of the thermal camera, we use a calibration target made of black-and-white melamine that looks like a checkerboard pattern in the thermal and RGB images. For the extrinsic calibration, we place a circular calibration target in the common field of view of the cameras and the LIDAR and compute the extrinsic parameters by minimizing an objective function that aligns the edges of the circular target in the LIDAR to its corresponding edges in the RGB and thermal images. We illustrate the convexity of the objective function and discuss the convergence of the algorithm. We then identify the various sources of coloring errors (after cross-calibration) as (a) noise in the LIDAR points, (b) error in the intrinsic parameters of the camera, (c) error in the translation parameters between the LIDAR and the camera and (d) error in the rotation parameters between the LIDAR and the camera. We analyze the contribution of these errors with respect to the coloring of a 3D point. We illustrate that these errors are related to the depth of the 3D point considered - with errors (a), (b), and (c) being inversely proportional to the depth, and error (d) being directly proportional to the depth.

KW - Cross-calibration

KW - RGB thermal mapping

KW - thermal cameras

UR - http://www.scopus.com/inward/record.url?scp=85027872315&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85027872315&partnerID=8YFLogxK

U2 - 10.1142/S2301385017500054

DO - 10.1142/S2301385017500054

M3 - Article

AN - SCOPUS:85027872315

VL - 5

SP - 59

EP - 78

JO - Unmanned Systems

JF - Unmanned Systems

SN - 2301-3850

IS - 2

ER -