| Abstract [eng] |
The final master’s thesis aims at developing and analysing a system for defect detection in material extrusion-based 3D printing process. Long manufacturing times so common in 3D printing, may lead to unnoticed defects that pose threat not only to the part being manufactured, but to the printer itself. In the first part of the thesis, we pick out defects that pose the greatest danger to printing equipment and discuss various methods for defect detection. The second part of the thesis focuses on adapting the system to specific 3D printer in question and explaining how template images were made. The third and final part consists of defect detection system analysis and parameter selection. The developed defect detection system is based on color segmentation and image subtraction methods for detecting zones with missing or too much material, and on edge detection and template matching for detecting layer shift defects. The efficiency of said system is 86,7% when used on parts made from different color material than the print bed. For using parts with similar color to the print bed, it is recommended to coat the print bed with a different color otherwise, system efficiency plummets to 65%. In any case, all experiments done on quality parts with no defects shows no false-positive detections. |
