Research On Micro Vision System For Automatic MEMS Wire Bonding

Wire bonding is the most extensively used packing technology for MEMS manufacturing. While the existing MEMS packing processes are usually operated by manual labor under microscopes, and the consistency and the finished product rate are highly influenced by operators'experience and mental state. With the development of MEMS industry, a higher request for automation wire bonding is put forward.In the whole bonding application, bond targets'identification and orientation are the most important aspects for full-automation. This dissertation introduces zoom microscope into the microscope positioning system to settle the contradiction between microscope's visual range and depth of field. Multiple scale images go through various image processing, and finally different bond targets are identified and positioned. We call this measure as"global-coarse and local-precise orientation", with which full-automation bonding is realized.First a micro vision positioning system based on zoom microscope is established after analysis of MEMS parts'property and bonding requirement, which gets both big visual range and high resolution. Global-coarse and local-precise orientation method has ensured system's efficiency and provide enough argument for development of full-automation.Sum-of-gradient-magnitude measure is chosen to adapt all images with different enlargement according to analysis of focusing measures'character. For 3D bond targets, it is explained that why focusing window should be placed accurately. Efficiency of the whole system is ensured by analyzing shift of focal plane when enlargement ratio is changing. Finally A novel algorithm flow is made to achieve 3D focusing with zoom microscope.For low enlargement images, global-coarse orientation is realized with auto-thresholding, corrosion-inflation operation and connected components labeling. And for large enlargement images, local-precise orientation is achieved by multiresolution wavelets segmentation on focus measure technique, which use the phenomena that images blur when out-of-focus. These two application...