An Adaptive Parts Counting Method based on Orthogonal Intensity Distribution Analysis for Industrial Vision Systems

Abstract

In this paper, we proposed a refined counting method of vibratory parts feeders for industrial vision system. Frames of feeding parts are continuously monitored by industrial camera at high speed, therefore efficient and accurate counting method is critical to adopt. Previous research converted two dimensional spatial signal of parts into a time series of one dimensional signal thus achieved efficient counting. In this paper, we further suppressed the counting error in previous research by also taking the variance of feeding speed of each parts into consideration meanwhile conserve its efficiency.