Development of Machine Vision System for Automatic Size Detection

Abstract

Everyday, logistic companies process millions of parcels for deliveries, transit movement, package sorting and storing. Automation of these processes will greatly reduce dependencies on human workers and increases efficiency of parcel handling. In order to automate package sorting and storing, a logistic company must also automate the measuring process of the package size via sensors or machine vision. This paper presents the development of a machine vision system for automatic box size detection using single camera vision and grid-less background. This system consists of a camera and a plain platform. The system combines canny edge detector algorithm for edge detection, Gaussian and Bilateral Filtration algorithm for pre-processing and statistical calculation for final dimension measurement. The system has 0.1 cm dimension resolution and achieved 94 % accuracy when determining width and height of boxes. The use of this single camera vision system may eliminate the use of physical detection sensors in box size measurement.