Wavelet-Based Denoising Framework to Enhance GAN-Generated Electrical Tree Images in High Voltage Insulation Analysis

This paper presents a novel wavelet-based denoising framework specifically designed for enhancing GANaugmented electrical tree images in high-voltage insulation analysis, a critical but challenging domain in power system diagnostics. The framework addresses the limitations of GANaugmented images by leveraging wavelet transform techniques to effectively suppress noise and artifacts while preserving the intricate branching patterns crucial for electrical tree analysis. Through multi-resolution analysis, the framework ensures improved visibility of fine structural details essential for accurate degradation assessment. Experimental results demonstrate the framework's superior performance over conventional methods, achieving higher PSNR (21.3070) and SNR (17.8930) values while maintaining lower RMSE (0.0861). The framework's effectiveness in preserving critical electrical tree features while reducing noise demonstrates its potential for advancing high-voltage insulation diagnostics, particularly in scenarios where accurate visualization of degradation patterns is essential for system reliability assessment.