Date of Award

8-2018

Degree Type

Thesis

Degree Name

Master of Science in Agricultural and Biological Engineering

Department

Agricultural and Biological Engineering

Committee Chair

Jian Jin

Committee Member 1

Dennis R. Buckmaster

Committee Member 2

Dharmendra Saraswat

Abstract

During recent years, hyperspectral imaging systems have been widely applied in the greenhouses for plant phenotyping purposes. Current imaging systems are mostly designed as either top view or side view imaging mode. Top-view is an ideal imaging angle for top leaves which are often more flat with more uniform reflectance. However, most bottom leaves are either blocked or shaded from top view. From side view, most leaves are viewable, and the entire structure can be imaged. However, at this angle most of the leaves are not facing the camera, which will impact the measurement quality. At the same time, there could be advantages with certain tilted imaging angle between top view and side view. Therefore, it’s important to explore the impact of different imaging angles to the phenotyping quality. For this purpose, we designed a swing hyperspectral imaging tower which enables us to rotate the camera and lighting source to capture images at any angle from side view (0◦) to top view (90◦). 36 corn plants were grown and divided into 3 different treatments: high nitrogen (N) and well-watered (control group), high N and drought-stressed, and low N and well-watered. Each plant was imaged at 7 different angles from 0◦ to 90◦ with an interval of 15◦. According to different treatments applied on experimental samples, two comparative pairs were set up: drought-stressed group vs. control group (Pair 1); N-deficiency group vs. control group (Pair 2). In this study, normalized difference vegetation index (NDVI) and relative water content (RWC) were computed and compared to determine optimized imaging angle(s). For NDVI, the imaging angle near to top view is optimized to separate Pair 1, while, the imaging angle near to side view is optimized to distinguish Pair 2. For RWC, partial least square regression (PLSR) models were applied to predict pixel-level RWC distribution of each plant, and higher imaging angles (close to top view) are better to tell the RWC distribution difference in Pair 1. In conclusion, higher imaging angles (close to top view) are better to separate different water treatments, while, lower imaging angles (close to side view) are better to separate different N treatments.

Share

COinS