<p>Radiometric quality assessment is essential in quality assurance and quality check processes of airborne laser scanning (ALS) data. Its roles in guaranteeing the precision of mapping products before and after flight missions become indispensable with the outsize reliance on the radiometric quality inference of multispectral ALS systems. An emerging theme in radiometric quality assessment is formulating a metric that preserves interpretability over multiple situations and hierarchical scales. Unfortunately, its progress coincides with a research community whose focus, until the recent development of intensity correction and radiometric calibration techniques, has been on geometric quality emphasizing system calibration and strip adjustment. Here, we propose a Boltzmann entropy-based multi-scale conceptual model incorporating data, model, and analysis scales for radiometric quality inference of multispectral ALS data after correction. Experimental work on three sets of Optech Titan data collected with different land cover scenarios was performed to justify Boltzmann entropy and the proposed framework. Experimental results reveal a reduction of stripe artifacts after radiometric correction, leading to improvements in classification accuracy by 0.3–3.3%. The Boltzmann entropy difference (<InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="41651_2025_213_Article_IEq1.gif" Format="GIF" Height="19" Rendition="HTML" Resolution="72" Type="Linedraw" Width="28" /> </InlineMediaObject> <EquationSource Format="TEX">$\Delta \mathcal {\hat{S}}$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi mathvariant="normal">Δ</mi> <mover accent="true"> <mi mathvariant="script">S</mi> <mo stretchy="false">^</mo> </mover> </mrow> </math></EquationSource> </InlineEquation>) before and after radiometric correction reveals the quality improvement difference over various channels. It highly correlates with the discrepancy-based distance (<InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="41651_2025_213_Article_IEq2.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="16" /> </InlineMediaObject> <EquationSource Format="TEX">$r^2$</EquationSource> <EquationSource Format="MATHML"><math> <msup> <mi>r</mi> <mn>2</mn> </msup> </math></EquationSource> </InlineEquation>=0.71) measuring classification/segmentation performance. <InlineEquation ID="IEq3"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="41651_2025_213_Article_IEq3.gif" Format="GIF" Height="23" Rendition="HTML" Resolution="72" Type="Linedraw" Width="41" /> </InlineMediaObject> <EquationSource Format="TEX">$|\Delta \mathcal {\hat{S}}|$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mrow> <mo stretchy="false">|</mo> <mi mathvariant="normal">Δ</mi> </mrow> <mover accent="true"> <mi mathvariant="script">S</mi> <mo stretchy="false">^</mo> </mover> <mrow> <mo stretchy="false">|</mo> </mrow> </mrow> </math></EquationSource> </InlineEquation> also gradually decreases as the resolution of intensity data is lowered, consistent with the coarse-graining of stripe artifacts. Overall, our work contributes the much-needed innovations to this nascent and overlooked field, and better equips ALS applications to capitalize on the rise in use cases of radiometric quality assessment techniques.</p>

Inferring Radiometric Quality of Multispectral Airborne Laser Scanning Data

期刊论文
发表日期：

2025年1月28日

查看全文

Xinghua Cheng,
Wai Yeung Yan

摘要

Radiometric quality assessment is essential in quality assurance and quality check processes of airborne laser scanning (ALS) data. Its roles in guaranteeing the precision of mapping products before and after flight missions become indispensable with the outsize reliance on the radiometric quality inference of multispectral ALS systems. An emerging theme in radiometric quality assessment is formulating a metric that preserves interpretability over multiple situations and hierarchical scales. Unfortunately, its progress coincides with a research community whose focus, until the recent development of intensity correction and radiometric calibration techniques, has been on geometric quality emphasizing system calibration and strip adjustment. Here, we propose a Boltzmann entropy-based multi-scale conceptual model incorporating data, model, and analysis scales for radiometric quality inference of multispectral ALS data after correction. Experimental work on three sets of Optech Titan data collected with different land cover scenarios was performed to justify Boltzmann entropy and the proposed framework. Experimental results reveal a reduction of stripe artifacts after radiometric correction, leading to improvements in classification accuracy by 0.3–3.3%. The Boltzmann entropy difference ( $\Delta \mathcal {\hat{S}}$ $Δ \hat{S}$ ) before and after radiometric correction reveals the quality improvement difference over various channels. It highly correlates with the discrepancy-based distance ( $r^2$ $r^{2}$ =0.71) measuring classification/segmentation performance. $|\Delta \mathcal {\hat{S}}|$ $| Δ \hat{S} |$ also gradually decreases as the resolution of intensity data is lowered, consistent with the coarse-graining of stripe artifacts. Overall, our work contributes the much-needed innovations to this nascent and overlooked field, and better equips ALS applications to capitalize on the rise in use cases of radiometric quality assessment techniques.