Defects Detection of Profile and Rail Surface with Image Processing Techniques

In the manufacturing sector and engineering, the system need for image processing with high performance capability is further increased every day. Image processing functions are used in many areas of medical and industry for product inspection, classification and also estimation of weather and agricultural productivity. Thus saving time and money in the development process and the introduction of new products in shorter period of time are achieved. During the production phase, fast detection of faulty or defective products will prevent defective parts to enter the market. In the hot rolling process, some defects such as cracks, shells and pits may be encountered. During metal forming process such as rail and profile, defects are checked by destructive or non-destructive inspection methods. Destructive inspection is not widely used because it creates damage on the material. Visual inspection and Eddy current commonly used as non-destructive methods. These methods are not useful because they are time consuming and having high error rates. Image processing techniques applied by computer on steel microstructure, mechanical behavior and surface structure, have been a subject of interest for last few years. In this study, a new algorithm is developed by using Watershed method to detect the surface defects which occur after the hot rolling process of profile and rail. This image pre-processing algorithm is applied to gray-level image obtained by a CMOS camera. At this stage, defective and nondefective image is separated and then defects areas are identified.