Measurement of section measurements in aluminum extrusion profiles with images in RGB channels

With the development of science and technology, manufacturability and quality control processes in the aluminum industry have become more complex. This brings with it some difficulties in production and quality control processes. The products desired to be produced contain thinner, fragile and complex structures. Traditional contact measurement techniques have become unable to respond to these and similar needs of the industry. However, with the use of image processing technologies in the industry, sensitive quality control processes have begun to be performed. In this study, it is aimed to perform quality control operations on profiles with complex structure and many measurement points, produced in aluminum extrusion profile production facilities, by using image processing techniques in a shorter time and with smaller margins of error compared to contact measurement techniques. In order to obtain the profile images for quality control measurements, the images obtained in the digital camera were converted into images in RGB channels and evaluated separately. These images were freed from noise after undergoing various filtering processes. Then, pixel-sized operations were performed on the image and measurements were made between the two specified points of the profile. By using this method, the negative effects of contact measuring devices are eliminated. By using image processing technique, measurements were made with a single device, in a very short time, with consistent numerical values. With the quality control process carried out with image processing, the necessity of giving the training of each of the measurement athletes used in the traditional method to the personnel has been eliminated. By reducing the margin of error, wrong productions were prevented. The measurement results were compared with the results of contact measuring devices such as calipers. The percentage error rates with the developed technique were compared with the contact measurement techniques and it was seen that the image in the B channel gave relatively more accurate results compared to the images in the R and G channels.