Inclusion Detection in Injection-Molded Parts with the Use of Edge Masking

Area of Science:

• Industrial Engineering
• Computer Vision
• Materials Science

Background:

• Plastic objects from injection molding often have flaws like inclusions, which are difficult to detect due to complex geometries and distracting edges.
• Current quality control methods struggle with accurate inclusion detection on irregularly shaped objects.

Purpose of the Study:

• To develop and validate an automated algorithm for detecting inclusions in plastic objects.
• To improve the accuracy and efficiency of quality control in plastic injection molding.

Main Methods:

• A novel algorithm classifies objects and uses edge masks to exclude irrelevant areas from analysis.
• Inclusion detection is performed by analyzing local variations in surface grayscale within unmasked image regions.
• The algorithm was tuned based on feedback from human quality controllers.

Main Results:

• The proposed method successfully detects inclusions, even with significant grayscale changes at object edges.
• Experiments on real-world rejected parts demonstrated high sensitivity without false positives at edges.
• The tuned algorithm achieved 100% recall and 87% precision, meeting industrial requirements.

Conclusions:

• The developed algorithm offers a robust solution for automated inclusion detection in plastic injection molding.
• This approach enhances quality control by providing high accuracy and reliability for complex object geometries.
• The system's performance is suitable for practical industrial applications, improving defect identification.