|
Wire-bond quality monitoring is a critical step in the integrated circuit packaging process to ensure the reliability of the wire connections. This paper introduces a highly effective algorithm for the reconstruction of wire structures, facilitating the rapid 3D reconstruction and parameter measurement of wire bonding structures. The procedure starts by capturing optical slice images of the wires using oblique-illumination optical sectioning microscopy. This technique restricts the axial emission range of structures within the microscope's field of view, thereby improving the axial resolution of the target structures. Subsequently, we employ a block-boundary-based point cloud segmentation algorithm to enhance the separation efficiency of wire structures. This algorithm divides the collected three-dimensional point cloud data into fixed-size blocks and applies parallel processing to reduce the necessary computation time for connectivity domain statistics. Additionally, parallel processing at the segmentation boundaries addresses the merging needs of point cloud label data, achieving rapid localization of connected structures. This methodology substantially improves the separation efficiency of the bonding structures and augments their utility in semiconductor packaging. Ultimately, high-precision and efficient 3D parameter measurement of wire bonding structures is achieved. Theoretical analysis and experimental results confirm the effectiveness of the proposed method.
|
