Error Control of Cylindrical Objects with Image Processing

Image processing methods have been widely used for detection of product faults in recent years. Images of industrial products that made of various materials are taken with different types of cameras and by using different lenses. Working on images of metallic products is difficult because of reflections, stains and halation on the object. Telecentric lens and telecentric illumination are used to come up with this problem. In this study, the axis shift due to the welding of a sphere and a cylindrical body was detected using image processing methods. An image was taken every 60 degrees when the object was rotated. The error of the product was detected on 6 images of each product. Firstly, the images that are colored are converted to black and white. Then the gaps in the image were destroyed by morphologic methods. Then positions of the two joined parts relative to each other are used. The position information of six views of an object is compared. According to the differences in relative location information, it is decided that object is produced with faulty or not. At the same time measurements of the size of the body were also made.