Side-Slither Data-Based Vignetting Correction of High-Resolution Spaceborne Camera with Optical Focal Plane Assembly

Chaochao Chen, Jun Pan, Mi Wang, Ying ZhuSensors 2018 · DOI

TL;DR

This paper introduces a novel method for correcting vignetting in high-resolution satellite imagery acquired by optical focal plane assemblies. By leveraging unique "side-slither" scan data, the approach effectively standardizes image data, identifies uniform regions using Gray-Level Co-occurrence Matrices (GLCMs), and applies a power-law model to correct for vignetting, achieving significantly improved image uniformity and quality.

What Problem Does This Paper Address?

Modern high-resolution optical satellite systems often employ "push-broom" imaging with large focal plane assemblies to capture wider swaths. A common challenge with these systems, particularly those using optical butting (where multiple sensors are optically joined), is vignetting. Vignetting causes a gradual decrease in image brightness towards the edges of the sensor's field of view, especially in the overlapping regions between individual sensors. This degradation significantly impacts image interpretability and requires robust correction methods. Traditional methods often rely on laboratory calibration or complex on-orbit procedures, which can be time-consuming or less effective under varying conditions. This paper aims to address the need for an accurate and efficient vignetting correction method that can be applied using readily available satellite data.

Methods

The proposed method consists of three main steps:

  1. Side-Slither Data Standardization: Raw side-slither data is first processed to ensure that each row exhibits consistent features. This involves basic adjustment and then enhanced adjustment using line detection (Sobel operator and Hough transform) to correct for slight misalignments and ensure accurate slope calculations. This step is crucial for treating the side-slither data as a uniform scene.
  2. Sample Point Extraction using GLCM: To identify uniform regions for calibration, the paper utilizes Gray-Level Co-occurrence Matrices (GLCMs) and the Inverse Difference Moment (IDM). GLCMs analyze spatial correlations of gray levels, providing texture information. By examining the GLCM and IDM in the standardized side-slither data, uniform regions are identified. Sample points are then extracted from these regions, representing the relationship between the reference (non-vignetting) area and the detector's response in the vignetting area.
  3. Power-Law Model Fitting: A power-law model is employed to fit the extracted sample points and derive vignetting correction coefficients. This model is chosen for its flexibility, being able to represent both linear and non-linear radiometric responses. The Levenberg-Marquardt (LM) algorithm is used to optimize the parameters of this power-law model. The resulting model is then used to correct the vignetting in the images.

Datasets

The study utilizes data from the YaogGan-26 optical remote-sensing satellite. The experimental data is divided into three groups:

  • Group A: Side-slither data used for calibrating and calculating the correction coefficients.
  • Group B: Another set of side-slither data used for verifying the calculated coefficients. This data is treated as a uniform field.
  • Group C: Data obtained using the classical normal push-broom mode, including scenes of water, city, hill, and desert. This group is used to verify the effectiveness of the correction on different types of ground targets.

Key Insights and Ideas

  • Side-Slither Data as a Uniform Scene: The core insight is that side-slither scan data, due to its nature of imaging the same target from multiple angles as the satellite moves, can be effectively treated as a uniform scene. This characteristic is leveraged to simplify and improve the vignetting correction process.
  • GLCM for Uniform Region Identification: Using GLCMs to analyze texture and spatial correlations is a novel approach for identifying uniform regions within potentially non-uniform satellite imagery, especially in the context of vignetting correction.
  • Power-Law Model Versatility: The power-law model's ability to adapt to both linear and non-linear radiometric responses makes it a robust choice for correcting vignetting, which can exhibit complex behavior across different illumination levels.
  • Improved Accuracy in Low-Brightness Regions: The proposed method demonstrates superior performance in low-brightness regions, where traditional methods often struggle, by effectively mitigating streaking and improving uniformity.

Key Results

The proposed side-slither data-based vignetting correction method demonstrated significant improvements:

  • Mean Value Change: Corrected images showed less than 1% change in mean value compared to raw data.
  • Root-Mean-Square Deviation (RA): Achieved RA values better than 0.1% for uniform scenes (Group B data), indicating high uniformity.
  • Streaking Metrics: Average streaking metrics were superior to 0.02, with maximum streaking metrics for the proposed method generally below 0.1, significantly outperforming other methods.
  • Qualitative Analysis: Visual inspection of corrected images across various scenes (water, city, hill, desert) showed better uniformity and reduced vignetting artifacts, particularly in low-brightness areas, compared to polynomial fitting, laboratory coefficients, and on-orbit coefficients.

Limitations and Future Work

  • Dependence on Side-Slither Data: The method's effectiveness is directly tied to the availability and quality of side-slither data for calibration.
  • Computational Cost: While effective, the GLCM analysis and power-law fitting might be computationally intensive, though the paper doesn't explicitly detail this.
  • Future Work: The paper suggests that further research could explore applying this method to other types of optical satellite sensors and investigate more advanced models for even greater accuracy. The authors also highlight the potential for this method to improve radiometric calibration in general.

Summary updated

Abstract

Optical focal plane assemblies are increasingly being used in high-resolution optical satellite systems to enhance the width of the image using linear push-broom imaging. With this system, vignetting occurs in the area of overlap, affecting image quality. In this paper, using the characteristics of the side-slither data, we propose side-slither data-based vignetting correction of a high-resolution spaceborne camera with an optical focal plane assembly. First, the raw side-slither data standardization is used to ensure that each row has the same features. Then, with the spatial correlation of a gray-level co-occurrence matrix, the gray-level co-occurrence matrix is proposed to identify the uniform regions, to extract the sample points. Finally, due to the characteristics of compatible linear response and non-linear response, the power-law model was used to fit, and the Levenberg⁻Marquardt algorithm was used to fit the model. In the experiment, polynomial fitting, laboratory coefficients and on-orbit coefficients were used for comparison with the proposed method. The side-slither data can be treated as a uniform scene due to their characteristics, and the side-slither image that was corrected using the proposed method showed less than 1% change in mean value, a root-mean-square deviation value better than 0.1%, and the average streaking metrics were superior to 0.02. The results showed that the proposed method performs significantly better in the vignetting area.
BibTeX
@inproceedings{chen2018side,
  author = {Chaochao Chen and Jun Pan and Mi Wang and Ying Zhu},
  title = {{Side-Slither Data-Based Vignetting Correction of High-Resolution Spaceborne Camera with Optical Focal Plane Assembly}},
  booktitle = {Sensors},
  year = {2018},
  doi = {10.3390/s18103402},
}
Figure