Abstract
An automatic surface defect detection system for hot ingot-casting-slabs has been developed to support hot-direct-rolling operations. Among several possibilities considered, the thermal image of a hot slab provided by an infrared scanning camera was evaluated as being the best. Having shown that the majority of surface defects could be recognized visually on the thermal image, an image processing system to replace visual recognition was developed. The image processing system as designed consists o f a signal processing unit and a mini-computer. Several different data processing stages are involved in the extraction of defect signals. Many parameters in the processing system were given a range of freedom and the final values adjusted to provide optimum detectability for actual defects. It was concluded that this detection system would enable on-line defect detection if some minor improvements were made. Two topics now under consideration are the coupling o f the detection system to an automated spot scarfer and the application of the system to defect detection in continuous-castingslabs.
