A JOINT CALIBRATION METHOD OF 3D IMAGES ACQUIRED BY COMMON OPTICAL PATH PAYLOAD

Geer Teng, Mei Zhou, Jian Hu, Haohao Wu, Wenqiang Li, Hanyuan Zhang, J. Y. Chen, Jie Wang2019 · DOI

TL;DR

This paper introduces a novel joint calibration method for "common optical path" payloads that simultaneously capture LiDAR point clouds and CCD images. By integrating ground static calibration with flight dynamics calibration, they significantly reduce ranging and misalignment errors, leading to a substantial improvement in the quality of 3D images.

What Problem Does This Paper Address?

Modern remote sensing often relies on combining data from different sensors to leverage their complementary strengths. Specifically, LiDAR excels at providing precise 3D structural information, while CCD cameras capture high-resolution textures. "Common optical path" payloads integrate these sensors into a single unit, simplifying data acquisition and initial registration. However, achieving high-precision 3D reconstructions requires meticulous calibration to account for inherent errors in both LiDAR (ranging errors) and the relative alignment between the LiDAR and CCD sensors (misalignment errors). Existing calibration methods often address these issues separately or are tailored to individual sensors, leaving a gap for a unified approach that optimizes the performance of these integrated payloads. This paper aims to bridge this gap by proposing a joint calibration strategy.

Methods

The proposed method is a multi-stage process designed to refine the accuracy of the common optical path payload:

  • Optical Axis Parallelism Calibration: This initial step ensures that the optical axes of the LiDAR and CCD sensors are perfectly parallel. This is achieved through indoor optical axis alignment debugging, where the coincidence of imaging points in the collimator focal plane for different parallel light angles indicates parallelism.
  • Ground Static Calibration: This phase focuses on correcting LiDAR ranging errors. It employs a "six-segment analytical method" using a specially designed detachable target with seven markers. By scanning these markers from seven positions and comparing the LiDAR measurements with total station data, additive and multiply constants for each laser receiving unit are calculated. This process refines the raw distance measurements.
  • Flight Dynamics Calibration: This is the core of the "joint" calibration. It addresses the misalignment errors between the LiDAR and CCD sensors. The process involves:
  • Improving the absolute accuracy of CCD image pixels using image control points.
  • Precisely locating ground-based feature points within the CCD images to identify corresponding LiDAR probes.
  • Leveraging the fixed matching relationship between CCD and LiDAR probes inherent to the common optical path.
  • Using ground-based LiDAR feature points to solve for LiDAR ranging and misalignment errors via a least squares method. This step integrates the refined LiDAR data with the accurate CCD imagery.

Datasets

The paper does not explicitly mention the use of publicly available datasets for training or evaluation. Instead, it describes a flight trial conducted in BaiMajing Town, Danzhou City, Hainan Province, China, using the common optical path payload. The data collected from this flight trial, specifically point clouds and CCD images from a validation area and a roof, were used for evaluating the calibration method.

Key Insights and Ideas

  • Synergy of Joint Calibration: The most significant insight is that a joint calibration approach, combining ground static and flight dynamics calibration, is more effective than addressing errors independently. This is because the errors are often interdependent in integrated payloads.
  • Importance of Optical Axis Alignment: The paper emphasizes that accurate optical axis parallelism is a prerequisite for effective joint calibration, forming the foundation for subsequent error reduction.
  • Specialized Target Design: The development of a detachable laser footprint target is a practical innovation that simplifies the identification of large laser footprints, which can be challenging at long distances.
  • Quantitative Improvement: The results demonstrate a clear and significant improvement in the accuracy of 3D data after the joint calibration, highlighting its practical value.

Key Results

The joint calibration method led to a substantial reduction in elevation residual errors:

  • Validation Area: The standard deviation of the elevation residual was improved from 0.54 m to 0.14 m.
  • Roof Area: The standard deviation of the elevation residual was improved from 1.83 m to 0.26 m.

These quantitative results, summarized in Table 2, indicate a significant enhancement in the accuracy and quality of the 3D images generated by the common optical path payload.

Limitations and Future Work

  • Limitations:
    • The paper focuses on a specific type of common optical path payload (linear LiDAR and linear CCD). Its applicability to other sensor configurations might require adaptation.
  • The calibration process involves both indoor and outdoor procedures, which can be time-consuming and require specific environmental conditions.
  • The paper does not extensively discuss the computational complexity or processing time of the proposed method.
  • Future Work:
    • The authors suggest that this method can effectively improve the quality of three-dimensional images, implying that further research could explore its application in more complex environments or for specific mapping tasks.
  • While the paper demonstrates effectiveness, further investigation into optimizing the calibration procedures for efficiency and robustness in various conditions would be beneficial.

Summary updated

Abstract

The common optical path payload is a new type of imaging payload that can acquire LiDAR data and CCD images simultaneously. This new payload integrates the linear LiDAR and linear CCD images according to the common optical system, and achieves the registration of LiDAR point cloud and CCD images by the alignment of common aperture optical axis and time synchronization control in the front of hardware. Based on the fixed matching relationship between CCD probes and LiDAR probes offered by optical path of common, this paper proposes a joint calibration method, which reduces the ranging error and misalignment error. And the results verify that this method can effectively improve the quality of the three-dimensional images. The standard deviations of validation area and roof are improved from 0.54 m to 0.14 m and 1.83 m to 0.26 m respectively.
BibTeX
@inproceedings{teng2019joint,
  author = {Geer Teng and Mei Zhou and Jian Hu and Haohao Wu and Wenqiang Li and Hanyuan Zhang and J. Y. Chen and Jie Wang},
  title = {{A JOINT CALIBRATION METHOD OF 3D IMAGES ACQUIRED BY COMMON OPTICAL PATH PAYLOAD}},
  booktitle = {˜The},
  year = {2019},
  doi = {10.5194/isprs-archives-xlii-2-w16-221-2019},
}
Figure