2-D Wavelet Transform In Ball Grid Array (BGA) Substrate Conduct Paths Inspection

This study presents a novel approach by using two-dimension Wavelet Transform (2-D WT) to localize the open and short boundary defect candidates on Ball Grid Array (BGA) substrate conduct paths. Once the potential defects are addressed, traditional inspection algorithms can be implemented for further analysis on these defect candidates. Therefore, the defect-detecting scope and inspection effort will be significantly reduced. The binary BGA substrate image is processed that shows only the boundaries of BGA substrate conduct paths, which is further decomposed directly by 2-D WT. Since the most of wavelet energy is clustered in the edges of image objects, the wavelet transform modulus sum (WTMS) of each edge pixel on BGA substrate conduct path boundaries is initially collected. By comparing the WTMS of an edge pixel with its WTMS on adjacent decomposition level, an edge pixel is determined to be a strongly irregular edge point (e.g. potential open or short defects) or not. Real BGA substrates with synthetic boundary defects are used as test samples to evaluate the performance of the proposed approach. Experimental results show that the proposed method is able to capture all the open and short defects on BGA substrate conduct paths without any missing error by appropriate across-level ratio and specified decomposition level.