论文阅读-Bias Compensation in a Rigorous Sensor Model and Rational Function Model for High-Resolution S
学术论文研读：严格传感器模型与有理函数模型的偏差补偿效果分析
论文标题:
This paper presents a Bias Correction Technique to address the challenges in a Precise Sensor Model and Rational Function Model for High-Fidelity Satellite Imagery. The proposed method focuses on mitigating systematic errors that arise from sensor calibration and environmental factors, ensuring accurate image representation. By integrating advanced algorithms, the system achieves superior image rectification, particularly beneficial for high-resolution applications where precise georeferencing is critical.
文章地址:
文章英文摘要:
This study introduces three distinct bias-corrected geometric correction techniques for high-resolution satellite imagery processing. The developed approaches incorporate a bias-corrected rigorous sensor model framework（RSM）operating within orbital space, a corresponding RSM tailored for image space applications, and an enhanced rational function model（RFM）designed for image domain corrections. Both RSM methodologies leverage on-board data combined with sensor-oriented RPCs extracted from imagery metadata. Experimental trials were conducted using QuickBird, WorldView-1 Basic images, and WorldView-2 Basic images as test datasets. The empirical findings reveal that implementing a zero-order polynomial function within an orbital-space-based RSM yields superior accuracy compared to other methods. When comparing image-space-based RSM and RFM implementations, both approaches exhibit comparable performance characteristics with a documented maximum discrepancy of less than 0.1 meters as measured by root-mean-square error analysis. These outcomes confirm that all proposed algorithms achieve sub-pixel accuracy across diverse imaging scenarios except for cases involving purely translational transformations within image space operations
摘要大意
本文开发了三种偏置补偿模型用于高分辨率卫星图像的空间校正工作。所述开发的模型主要包括：轨道空间中基于严格传感器模型（RSM）实现的空间偏差校正；以及在图像空间中分别采用严格传感器RPC参数和有理函数模型（RFM）进行的空间偏差校正处理。其中，
RSM
与
RFM
均利用图元数据所提供的地面控制点坐标信息以及面向传感器的地物特征编码系数来进行精确的空间校正计算。
测试所选样本包含QuickBird、WorldView-1及WorldView-2等典型遥感影像数据集。
实验结果显示，在轨道空间内基于零阶多项式函数实现的空间偏差校正是提高校正精度的有效手段；对比分析显示，
RSM
与
RFM
在不同空间域内的表现具有较高的一致性，
RFM
算法所能达到的最大残差误差不超过0.1米。
进一步分析发现，
无论是在平面方向还是垂直方向上，
所有提出的算法均能显著提升遥感影像的空间定位精度。
补偿方法
针对共线成像模型
针对RPC模型
实验结果
实验研究证实，在应用偏置补偿型RSM方法时所获得的结果优于RFM方法。
个人理解
对两种补偿方案在精度上表现相似程度相近，在解决其他部分问题时,RPC所采用的像方仿射模型相较于共线方程具有更为简便易行的优势，并且能够方便地获取结果.然而,RPC模型中所包含的六个仿射参数在实际解算过程中由于存在较强的矩阵相关性和较低的矩阵可逆性等因素而导致计算过程中容易出现算法发散的现象,这方面的研究仍具有较大的探索价值.