Defect Atlas and Image Analysis of Surface Defects on CR Steel Sheets

The work presents a preliminary approch towards building up of an automated surface inspection system for recognizing defects on Cold Rolled Sheet. The sy- stemuses an area scan camera (as opposed to line-scan)for better image quality. the whole task has been divided into stages:

1. Building a preliminary Defect Atlas and
2. Formulation an approch towards classification of two simple defects (Hole and Pinch Mark).

In the first stage the images of defects were acquired on-site at Bokaro Steel Limited, and then arranged into an Atlas along with their image-acquisit- ion data (such as type of Illumination, light source and camera angles etc.). The Atlas (presented in Appendix) is constantly utilized to improve the Image- Acquisition procedure after assessing quality of defect images subjectively.

The second comprises of

1. Defect Modelling tacking the images from the Atlas, and
2. Devising aDefect Classification alogrithm, so that the system could distingwish atleast two surface defects from any set of images fed to the system.

This is a preliminary system for handling a very complex problem. The shortcomings in the procedure and suggestive approaches to overcome them are discussed .THe recent technological trends adopted by leading Surface Inspection System manufactures are also presented.

[Thesis List] [ME Home Page]