An application of information fusion to detect and measure the steel rods of online sensed images

This paper presents a method of on-line measurement of the amount of steel rods with the real-time image acquired from the production line site. The information fusion is used in this study, which includes preliminary knowledge about the shape and site of the objects, data and information about edges of objects detected using different operators, and the distributed features of the rod from "edge to center" abstracted with artificial neural networks. Fuzzy classifier, artificial neural network and fuzzy logic methods have been used as the algorithms of the fusion in different levels. Experiments indicated that fusion with appropriate information deliberatively chosen is able to achieve excellent performance with only very small set of data and features being used.