Vision Based Pattern Recognition For Industrial Inspection

Marizan, Sulaiman and Hairol Nizam, Mohd Shah and Lim, Wee Teck (2010) Vision Based Pattern Recognition For Industrial Inspection. Project Report. UTeM, Melaka, Malaysia. (Submitted)

PDF (Full Text) Vision_Based_Pattern_Recognition_For_Industrial_Inspection.pdf - Submitted Version Restricted to Registered users only Download (11MB)

Abstract

A Flexible Manufacturing System (FMS) is an adaptive and dynamic system that cater to a wide range of dllferent jobs where each involves a set of operations that are required to be done at a predetennined workstation. Therefore this report is regarding the application of a vision system develop in a previous project to monitor and control the operations of the system in order to manage the decision making concerning jobs and work pieces recognition that are to be made during system operation in real time. The vision algorithm will be applied in various different experiments and conditions to test and validate its efficiency and practicality. The vision system uses an edge detection method and template matching to achieve pattern recognition capabilities. In tllis report stress the experiments done, its results and its discussions. Firstly, five experiments were done from simple basic shapes, complex shapes, to barcode, alphabets and numbers. These experiments are able to test the vision algorithm on accuracy of matching, ease of usage during training phase and capability to modify and flexibility in determining the best features. User define solutions and open 1ended architecture of the program are the advantages oftl1is vision algorithm. From the results o:fthe experiments, an average of 92.75% for accepting Acceptable object and 100% for rejecting Rejected object were obtained, this results are consider acceptable and confirms the validity of the vision algorithm, ~hich is the detection of required features and not the detection of faulty features. For the discussion part it mostly discuss about the best way and setting to improve efficiency of the algorithm and also some of the limitation in this algorithm.

Actions (login required)

View Item