This paper introduces an innovative approach using AI (Artificial Intelligence) and wavelet-based techniques for creating high-detail mosaics of microscopic images. The approach is designed to overcome inherent challenges in alignment, illumination, noise, and the efficient processing of large datasets. Unlike prior methods that heavily rely on spatial analysis, the proposed method leverages wavelets to decompose images into different frequencies and employs an AI-based self-attention algorithm to preserve the important details in the stitched image. This novel approach excels in addressing various challenges, including noise reduction, accommodation of varying exposure levels, maintaining consistent image quality, and reducing computational time. These features collectively make our method a robust and reliable solution, particularly suited for defect characterization in critical domains such as nuclear and aerospace materials.