Development of an effective algorithm for microchip lead inspection

Abstract

A new effective method for the microchip lead inspection for the chip manufacturing industry has been developed in this research. This new technique is implemented using Lab View programming environment and tested on many sample images using National Instruments frame grabbers and hardware interfaces. The first part of the software deals with the image enhancement and auto-thresholding method to prepare for precision measurements. The second part of the software is using newly developed statistical analysis technique for automatic IC leads defect detection (such as linearity, implied planarity, offset...etc). Finally the research carries out a precision, repeatability and reproducibility test procedures of the system. Contrary to the gray scale pattern matching technique, the proposed technique employs selected parameters of binary blobs to perform fault detection and measurements. This leads to a significant reduction of image processing time. A special combination of gray level filtering techniques with gray morphological operations has been used to enhance the borders of the lead images. A newly developed threshold calibration technique significantly improves the measurement accuracy. A unique statistical analysis has been developed to identify all possible lead defects in the chips. This method is rotationally and scale invariant and able to detect defective leads for the chips with different specifications. The minimum required information about the microchip is the number of leads.