Thermal image enhancement for adverse weather scenarios: a wavelet transform and histogram clipping approach

Thermal imaging is crucial in surveillance, transportation safety, defense, search and rescue, industrial maintenance, manufacturing quality control, and firefighting for enhanced situational awareness. Yet, its effectiveness diminishes in adverse weather like fog, haze, rain, and poor lighting. However, its efficacy is often compromised in challenging environments characterized by adverse weather conditions such as fog, haze, rain, and poor lighting. This paper proposes an image enhancement framework using Wavelet Transform to improve thermal imaging in challenging conditions. It decomposes images into frequency components, selectively enhancing low and high-frequency details. A kurtosis-based histogram clipping algorithm enhances contrast and visibility in low-frequency components. Synthesizing components using inverse discrete wavelet transform produces the final image. The proposed methodology is evaluated using publicly available datasets, including OSU thermal and AAU rain-snow datasets, to evaluate its performance in various challenging scenarios with quantitative metrics like Entropy, Absolute Mean Brightness Error (AMBE), Peak Signal Noise Ratio (PSNR), and Structural Similarity Index (SSIM) are employed for a comprehensive analysis of the proposed Wavelet Transform-based approach. The results demonstrate image quality and visibility improvements, positioning the proposed framework as a promising solution for enhancing thermal imaging in challenging environments and extending its applicability in critical real-world scenarios.