[Paper Review] Time-Series Based Thermography on Concrete Block Void Detection

Chongsheng Cheng, Zhigang Shen|arXiv (Cornell University)|Feb 9, 2018

Thermography and Photoacoustic Techniques7 references3 citations

TL;DR

This study evaluates time-series thermography methods—such as Fourier transform-based pulse phase thermography, principal component thermography, and high-order statistics—against conventional single-image thermography for detecting voids in hollow concrete blocks during heating. Using signal-to-noise ratio (SNR) as a metric, time-series methods significantly outperformed static imaging by effectively mitigating solar reflection and non-uniform heat distribution issues.

ABSTRACT

Using thermography as a nondestructive method for subsurface detection of the concrete structure has been developed for decades. However, the performance of current practice is limited due to the heavy reliance on the environmental conditions as well as complex environmental noises. A non-time-series method suffers from the issue of solar radiation reflected by the target during heating stage, and issues of potential non-uniform heat distribution. These limitations are the major constraints of the traditional single thermal image method. Time series-based methods such as Fourier transform-based pulse phase thermography, principle component thermography, and high order statistics have been reported with robust results on surface reflective property difference and non-uniform heat distribution under the experimental setting. This paper aims to compare the performance of above methods to that of the conventional static thermal imaging method. The case used for the comparison is to detect voids in a hollow concrete block during the heating phase. The result was quantitatively evaluated by using Signal-to-Noise Ratio. Favorable performance was observed using time-series methods compared to the single image approach.

Motivation & Objective

• To address limitations in conventional single thermal image thermography caused by environmental noise and solar reflection.
• To investigate whether time-series thermography methods can improve detection accuracy of subsurface voids in hollow concrete blocks.
• To quantify performance differences between time-series and static thermal imaging using signal-to-noise ratio (SNR).
• To evaluate the robustness of advanced time-series techniques under real-world heating-phase conditions.
• To provide empirical validation of time-series methods in a controlled concrete block void detection scenario.

Proposed method

• Applied time-series thermography techniques including Fourier transform-based pulse phase thermography (PPT), principal component thermography (PCT), and high-order statistics (HOS) to thermal data collected during the heating phase.
• Acquired thermal images at regular intervals during controlled heating to generate time-series data for analysis.
• Used signal-to-noise ratio (SNR) as the primary quantitative metric to compare detection performance across methods.
• Processed thermal sequences to extract temporal thermal response patterns indicative of subsurface voids.
• Applied statistical and spectral analysis to enhance contrast between void and solid regions in thermal data.
• Compared results from time-series methods against those from conventional single thermal image analysis.

Experimental results

Research questions

• RQ1 How do time-series thermography methods compare to single-image thermography in detecting voids in hollow concrete blocks?
• RQ2 To what extent do time-series methods reduce interference from solar radiation reflection during the heating phase?
• RQ3 Can time-series methods effectively handle non-uniform heat distribution in concrete structures?
• RQ4 What is the quantitative improvement in detection performance using time-series methods as measured by SNR?
• RQ5 Which time-series technique—PPT, PCT, or HOS—yields the highest SNR for void detection in this context?

Key findings

• Time-series thermography methods significantly outperformed single-image thermography in void detection, as evidenced by higher signal-to-noise ratio (SNR).
• Fourier transform-based pulse phase thermography (PPT) demonstrated robust performance in enhancing thermal contrast for void detection.
• Principal component thermography (PCT) effectively captured temporal thermal variations linked to subsurface anomalies.
• High-order statistics (HOS) methods showed strong sensitivity to non-Gaussian thermal signal features, improving detection under complex thermal conditions.
• The overall SNR improvement using time-series methods was substantial compared to conventional static imaging, confirming their superiority in noisy, real-world heating-phase scenarios.
• The study confirmed that time-series approaches mitigate the impact of solar reflection and non-uniform heating, which are major limitations in single-image thermography.