Utilization and Configuration of a Multi-Spectral Image Sensor for Turfgrass Data Collection

The ultimate goal for golf course management is to generate high quality playing conditions based in sound economics and environmental quality. Meeting these goals often requires a site manager to implement a systematic and detailed process of data collection and analysis as the basis for management decisions. Sensors are a potentially viable data collection tool to meet these needs. One type of sensor, spectral reflectance sensors, measures reflected light from turfgrass leaves. The primary hypothesis of this research is that image sensors can be used as a viable data collection tool for physical and chemical turfgrass data and that image sensors provide an advantage of over sensing technologies or methods. Key advantages to image sensors include the ability to segment parts of the image and gain a more realistic depiction of site characteristics. The primary objective of this research is to complete foundational research that defines use and capabilities of image sensor for collecting detailed site data in turf. Research objectives are to (a) determine the effect of external parameters on utilization and configuration of a multi-spectral image sensor for turfgrass data collection, (b) determine effects of ambient light on image sensors for turfgrass data collection, (c) determine effects of sensor and image resolution on sensor output, and (d) advantages of image segmentation in image analysis. A multi-spectral image sensor was tested for capabilities to collect data used for reflectance calculations under different environmental conditions and sensor configurations. Environmental parameters and different sensor configurations did not significantly affect sensor output. Though, qualitative analysis of different environmental parameters and sensor configurations did suggest advantages to some sensor configurations. Further study demonstrated that a compensation model for the variable effects of ambient illumination could correct turfgrass data and produce horizontally linear reflectance responses on red, green, and near-infrared channels with lower standard deviations than have been produced in other studies of this model applied to agricultural crops. No significant trends in sensor output were identified with study of different sensor and image resolutions. Statistically significant advantages of image segmentation and thresholding were noted in this study.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Natural Resrouces and Environmental Sciences

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2015

Author dc:creator

• Schmidt, Mark Alvin

Contributors dc:contributor

• Tom Fermanian

• Engineering, Agricultural

Language dc:language

eng

Identifier

(MiAaPQ)AAI3160950

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/83102