An efficient approach for vision inspection of IC chips

The research aims to develop an automated vision inspection system of IC chips that used to detect the defects of marking and design shape of IC chips.As a result of higher failure probability of manual inspection system, this automated system is developed.The automated vision system will consists of five main phases which are image acquisition, image enhancement,image,segmentation, comparison on features and decision making.The features will be extracted from the target image using projection profile method.The decision will be made using the trained neural network to identify the four common defects of IC chips which are illegible marking,upside down marking and missing character on chip. The results of the automated system are to determine whether to accept or reject the chip. The results are computed within 10 seconds and have a high percentage of defects detection which is about 95 %. Through the results obtained, the automated vision inspection system of IC chips can be utilized in the manufacturing field to replace the manual inspection system.It can replace about five to eight inspection experts to reduce the cost of hiring and resources in about 70%. Other than that, the rate of accuracy and efficiency of detecting the defects are improved by 95% because the consistency of inspecting the chips is maintained from having variations of judgments by the experts