Off-campus South Dakota State University users: To download campus access theses, please use the following link to log into our proxy server with your South Dakota State University ID and password.

Non-South Dakota State University users: Please talk to your librarian about requesting this thesis through interlibrary loan.

Document Type

Thesis - University Access Only

Award Date

2000

Degree Name

Master of Science (MS)

Department / School

Agricultural Engineering

First Advisor

Daniel S. Humburg

Abstract

A machine vision system was designed, built and calibrated to locate mushrooms on a tray for harvesting. This system was designed to reduce the manual labor currently utilized in mushroom industry for harvesting. The system was built using commercially available off-the-shelf components with simple design for construction. Images of the mushrooms were acquired from two cameras to compute three dimensional physical attributes. The overhead (first) image was processed by software routines to isolate and identify the approximate centers ([x, y] coordinates) of individual mushrooms. This was followed by an edge tracing routine to obtain diameters of the mushroom heads. The inclined (second) camera was used in conjunction with structured lighting to extract the [x, z] coordinates in order to obtain the heights of the mushroom. Cameras had to be calibrated to predict the geometrical coordinates from the image coordinates. Several calibration techniques were reviewed, and a Direct Linear Transformation calibration routine was adopted for this work. Two calibration fixtures were utilized in the calibration. Error patterns were modeled to allow reduction of prediction errors. The overhead camera prediction errors were reduced from a maximum of 11 pixels to 2.15 pixels. The inclined camera prediction errors along the X and Z directions were independently considered for error reduction. The X direction errors varied as a cubic function with maximum being ±4.0 pixels. This error was reduced to ±1.40 pixels after applying the models. The Z direction errors were also reduced from 1.93 pixels to 0.66 pixels by developing error models.

Library of Congress Subject Headings

Mushrooms -- Selection -- Automation
Mushrooms -- Harvesting -- Automation
Computer vision
Image processing -- Digital techniques

Format

application/pdf

Publisher

South Dakota State University

Download

Share

COinS