Electronic Thesis and Dissertation Repository

Thesis Format

Monograph

Degree

Master of Engineering Science

Program

Electrical and Computer Engineering

Supervisor

McIsaac, Ken

Abstract

Plant height is a key phenotypic attribute that directly represents how well a plant grows. It can also be a useful parameter in computing other important features such as yield and biomass. As the number of greenhouses increase, the traditional method of measuring plant height requires more time and labor, which increases demand for developing a reliable and affordable method to perform automated height measurements of plants. This research is aimed to develop a solution to automatically measure plant height in greenhouses using low cost sensors and computer vision techniques. For this purpose, the performance of various depth sensing technologies was compared by considering the following: camera price, measurement resolution, the field of view and compatibility with the application requirements. After analyzing the alternatives, the decision was to use the Intel RealSense D435 3D Active IR Stereo Depth Camera. The algorithms developed were used to monitor plant growth of basil. Results demonstrated a promising performance of the developed system in practice.

Summary for Lay Audience

The decision on whether to irrigate is usually an automatic process in the greenhouses. However, growers might also decide on the allocation of water to plants based on their insights and preferences. A smart irrigation system can be used to plan and schedule watering periods at least as well as an experienced grower, which in turn can potentially improve the yield and reduce costs for many greenhouses. Among growth parameters, plant height can be used to detect water stress in smart irrigation applications. Measuring height of plants manually with a measuring tape or other handheld instruments may be inaccurate, time-consuming and can easily damage plants during the measurement process. The automation of plant height measurement can make this process more efficient, accurate, and suitable for large scale trials. 3D computer vision enables computers to perceive depth in digital images and generate a three-dimensional dataset from the scene. There are many 3D sensors that can provide depth information of the scene. However, some of them cannot work properly in the greenhouses due to the complexity of the greenhouse environments and the complexity of the plant itself. This study presents an automated and non-invasive method for accurate crop plant height measurement using an Intel RealSense D435 depth sensor. RealSense D435 is based on active infrared stereo technology which is a combination of various sensing technologies and can function effectively under daylight and lowlight conditions at a reasonable price. To start, data captured from the camera was preprocessed to remove invalid and undesired areas from the depth information. In the next step, it was required to separate the plant from the background (segmentation). In this study, segmentation was done by assuming that plants are green objects in the scene and are in a specific distance range from the camera. Finally, a technique was used for estimating plant height by finding the highest point of plants. The robustness of the proposed method to various changes in the environment was tested on five basil pots in two different environments. Plant height estimation results demonstrated high correlations and low average errors between estimated and actual values.

Download

Share

COinS