Identification and Inspection of Silverware Using Machine Vision

Abstract

This study investigates the identification and inspection of silverware in commercial dishwashing applications. The purpose of this study was to investigate methods to efficiently identify and inspect singulated silverware pieces. The major challenges encountered were limited processing time and poor image quality due to vibration of silverware pieces, specular reflections, ill-lighted regions and noisy power sources. It was desired to have a high processing rate with a reliable identification and inspection in place. The method was primarily experimental. Lighting equipment was designed to eliminate noise and to reduce specular reflections and ill-lighted regions. A vision system algorithm with a reliable identification and inspection procedure has been designed, developed and tested. Experimental results with 200 silverware pieces in each of 2 different silverware patterns demonstrate identification accuracy of 100% and classification accuracy of 86% for clean silverware pieces and 91% for dirty silverware pieces.