A feature extraction method for galvanized steel sheet powdering rates classification

Abstract

This paper describes a feature extraction method for classifying galvanized steel sheet powdering rates. By combining machine vision with statistics, and using the standard deviation's difference between powdered and normal regions, the range of powdered areas is recognized. The method exhibits a number of interesting features: it uses mathematical statistics for image feature analysis, and develops an effective method for analyzing particle size which cannot be measured by manual detection. The experiment result shows that the correction rate of this method to acquire galvanized steel sheet powdered regions is up to 99%, which satisfies the requirements of the application.

Keywords:

• MACHINE VISION
• FEATURE EXTRACTION
• SEED FILLING
• PARTICLE ANALYSIS