Defect Detection in Pattern Texture Analysis

The detection of abnormalities is a very challenging problem in computer vision, especially if these abnormalities must be detected in images of textured surfaces. The cast extrusion manufacturing process is the initial step which enables the creation of the raw materials, such as clear polypropylene film, needed for the flexible packaging manufacturing process. The current methodology of controlling extrusion related defects occurrences is attempted by a combination of statistical Sampling and human inspection. However, the defects are small in size and hard to visualize in a clear thin film 3 m in width moving at a speed of 50m/min/. This resulted in poor product quality and high return ratio from customers. This problem becomes even more complicated in case when testing is possible only with one surface of the part. Texture analysis plays an important role in the automated visual inspection of texture images to detect their defects. This automated classification method helps us to acquire knowledge about the pattern of defect within a very short period of time. This research investigates possible defect detection methodologies and has subsequently proposed a system that is capable of real time monitoring of defects on the cast extrusion manufacturing process.