Research on image quality optimization and point cloud processing algorithms for micro-bump height measurement

The micro-bumps on a wafer play a crucial role in both structural support and electrical connections between chips. Accurate measurement of micro-bump height is a critical step in advanced semiconductor packaging. An adaptive light intensity adjustment technique is proposed to automatically regulate the light source intensity, thereby providing high-quality images that facilitate key feature extraction. This technique enhances existing mainstream micro-bump measurement methods, which rely on feature extraction from micro-bump images. Additionally, a point cloud semantic segmentation method is proposed to enable rapid automated measurements. A Bruker standard sample and an optical flat were employed to evaluate the measurement precision of the developed system. The system demonstrated a repeatability precision of 0.04 μm, with a relative measurement error below 0.1%. Experimental results confirm that the proposed method is suitable for in situ micro-bump height measurement.