JESS: Re: Use of Jess in crack detection

[S F Palu]

Yes , this has always been an area of AI but I am not sure if anyone has used JESS in such area. I am no expert on the subject but I believe , a good starting point for you is to go to the following web site and look up some published papers under the subject of "Computer Vision" :   Scientific Literature Digital Library: ------------------------------------------------   http://citeseer.nj.nec.com/cs     I have come across some print publications (books) also on the internet relating to : "Automatic Visual Inspection Expert Systems" which have been deployed in the manufacturing industry. One system has been  developed specifically for the "silicon wafer fabrication". The architecture  used were a hybrids of  Support Vector Machines (SVM) , Artificial Neural Network (ANN) and Fuzzy Logic all working simultaneously. It used high speed digital cameras to capture the image of the component (chips etc) in along the production line , it then try to pattern match with known defect images from the database that were catalogue and classified previously. If there is a match then the component will not pass to the next stage of testing.   I believe that Bob Orchard from NRC (author of Java Fuzzy Logic Toolkit, FuzzyJ) who is on this mailing list, can probably point you out to more publications on using Fuzzy Logic for image analysis. Doing AI in imaging expert systems is one thing, but I think that you still need to dig deep into the area of "Digital Image Processing". The JAI  (Java Advanced Imaging) is an a well written API with a vast number of different "imaging operator routines" (filters)  implemented and also it is flexible which you can extend the standard operators with your own algorithm. I am impressed with JAI is that the lead spec for the JAI   JSR ( java specification request)  is planning to implement a range of  "wavelet operators" in its next version. This would even make developing "Automatic Visual Inspection Expert Systems" in Java more sophisticated.   I would recommend the following title as a good start:   "Digital Imaging Processing", 2nd Ed,  by  R.C.Gonzales & R.E.Woods published by: Prentice Hall.   The book is not Java based (although there are lots of Java based books on imaging at Amazon), but the last 3 chapters deals with Figure Recognition  and Image morphology which these are the areas of most relevant to what you are trying to do. All the image operators mentioned in this book are already available in  JAI, but JAI does not have any algorithm in pattern recognition which is something you have to implement it yourself.  The book has also listed about  160 references to books and journals in advance image analysis and intelligent image recognition systems.   Finally, I know there are people from the US military industries in this mailing list (JESS users)  who will probably give you advise or guides of where to look. This sort of technology has been used by the military way before civilian commercial applications started appearing on the market. Such use is intelligent imaging systems for picking ground target identification.   Cheers, Sione.       ----- Original Message ----- From: David To: JESS Sent: Saturday, October 11, 2003 5:04 AM Subject: JESS: Use of Jess in crack detection Hi guys,   I am thinking of using an AI system for crack detection. Basically, I am doing some work in silicon wafer fabrication defect detection. The data will be input from an source where it will resembles a crack. I am trying to use some AI in this crack detection scheme. In the past, crack detection is entirely done using human eyes. I wonder how feasible it is to use Jess as an AI engine to identify these cracks from a visual image.   Many thanks in advance for your time and attention.   Regards David