|Reese, Daniel - UNIV OF MD, COLLEGE PARK|
|Lo, Martin - PROF. UMCP|
|Narayanan, Priya - UNIV OF MD, COLLEGE PARK|
Submitted to: International Food Technology
Publication Type: Abstract Only
Publication Acceptance Date: July 15, 2007
Publication Date: July 28, 2007
Citation: Reese, D.Y., Lefcourt, A.M., Kim, M.S., Lo, M., Narayanan, P. 2007. Whole surface image reconstruction for machine vision inspection of apples [abstract]. International Food Technology (IFT) Expo 2007. Technical Abstract: Recent outbreaks of E. coli O157:H7 in the US serve as reminders of food safety in produce. Unpasteurized apple juice/cider has been identified as a repeated source of E. coli O157:H7 contamination. Apples with diseased or fungal contaminated surfaces, and open skin cuts and bruises may become sites for bacterial growth. Developing detection technologies for apple defects and contaminations in the post-harvest preprocessing stage is therefore important for quality and especially for safety. Existing inspection systems using imaging techniques, however, are only capable of viewing half of the surface. The objective was to develop a method for whole surface imaging of apples using an RGB camera with real-time video transmission to a computer for data analysis. Two 200×235 mm mirrors (4-6 wpi with reflection coating > 90%) were mounted at various degrees and distances from a 3.25-inch weighted plastic apple. The apple was suspended 15.5 cm above a tabletop on parallel thin wires to allow maximum view of the apple. A color RGB camera was mounted directly above the apple. Whole surface images of the highest resolution were achieved with the mirrors located at 13.2 cm above the tabletop, 2.5 cm below the apple, and 6.3 cm away from the center of the apple to avoid image blockage. The mirrors angled at 33.4º gave the most coverage and best resolution of the bottom of the apple. The 3-D surface was reconstructed from the projected mirror images integrating voxels, a volume element representing a value on a regular grid in three dimensional space, and coordinates based on the area, aspect ratio, max./min. diameters, perimeter, and color attributes. Whole surface imaging of apples provides a novel, noninvasive, and economical method for detecting defects over the entire surface of an apple and allows better detection of sites susceptible to microbial contamination.