An Automated Planar Segmentation Engine for Precision Machine-Vision Quality Control

International Journal of Computer Science and Technology Vol. 1, No. 9, September 1999 | pp. 70–78

DOI: 10.46882/1999/IJCST/000009

Research Article

Title: An Automated Planar Segmentation Engine for Precision Machine-Vision Quality Control

Names of Authors: Hans Schmidt¹, Li Wei-Dong²

Authors’ Affiliations: ¹Department of Automation Technology, University of Stuttgart, Germany; ²Institute of Microelectronics, National Taiwan University, Taipei, Taiwan.

Abstract:
Microscopic surface anomalies on physical manufacturing plates introduce structural degradation risks, making highly precise automated visual sorting systems essential during production runs. Manual inspection workflows are time-consuming and error-prone, whereas traditional edge-detection tools struggle to categorize complex, layered material imperfections. This study details a custom image segmentation architecture optimized for real-time defect classification and geometric tracking on metallic plates. The model combines 14 structural processing layers with advanced spatial poolers to track defects across variable resolutions. We trained the vision model on an industrial manufacturing archive of 45,000 high-resolution optical inspection images. The developed engine recorded a mean Average Precision (mAP) metric of 96.5% across eight unique defect definitions, including microscopic cracks, superficial abrasions, and structural alignment flaws. This marks a 7.8% enhancement over classical baseline segmentation classifiers. The computation step averaged 12 milliseconds per verification image, comfortably supporting the rapid performance metrics demanded by modern automated production lines.

Keywords: Visual Inspection, Image Segmentation, Surface Defect Detection, Industrial Automation, Pattern Recognition, Quality Control.

Manuscript Timeline: Received: June 09, 1999; Revised: July 18, 1999; Accepted: August 10, 1999; Published: September 22, 1999.

Citation: Schmidt, H., & Wei-Dong, L. (1999). An Automated Planar Segmentation Engine for Precision Machine-Vision Quality Control. International Journal of Computer Science and Technology, 1(9), 70–78.

Subscribe to read the full article: https://internationalscholarsjournals.org/subscribe-to-read