Automated Quality Inspection On Tile Border Detection Using Vision System

Yap, Sher Kee (2019) Automated Quality Inspection On Tile Border Detection Using Vision System. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (24 Pages) Automated Quality Inspection On Tile Border Detection Using Vision System.pdf - Submitted Version Download (1MB)

Abstract

Most of the ceramic tile industry still doing the quality control by manually. The accuracy of the manual inspection by human is lower due to the harsh industrial environment with noise, extreme temperature and humidity. A camera should replace the human eyes to recognise the defect tile effectively. Thus, a suitable method have to investigate for implementing this function. This project aim to design and develop an automated quality inspection in ceramic tile industry using vision system. The performance in term of detection accuracy for the system is analysed. An Imaging Source Series CMOS industrial camera is used to capture the tile border. The system is implemented in the MATLAB software. Image processing with Canny edge detection technique and morphological operation are used to segment and extract the tile border edge. The threshold level of image processing, focus and iris of camera and illumination of the light are adjusted to improve the performance of the system. The system developed is only to detect cracks occur on the edge of the tile border, middle crack such as pinhole is not included. The overall automation process involves image capturing, image processing, defect detection algorithms and decision making. The defect detection algorithms are developed to differentiate the defective tile. The automated quality inspection for the defect detection of tile border using vision system based on the background subtraction method and gradient variation of the tile border edge are presented in this research. The system using background subtraction method has achieved 50% accuracy in identify the status of tile since it consist of many limitation. By evaluate the gradient variation on the tile border edge, the accuracy of the defect detection has achieved 80% in identify the tile condition. The performance of the second method is relatively strong since the process of detection is considered in many aspects. In future, a consistent workspace such as in a production line can be achieved and reduce the error. The good and defect tiles will be classified and divided to different place by design a conveyer sorting system.

Actions (login required)

View Item