Computer Science and Engineering, Department of


First Advisor

Ashok Samal

Second Advisor

James Schnable

Date of this Version


Document Type



Bashyam, S. (2016). Determination of Plant Architecture and Component Phenotyping Based on Time-lapse Image Analysis (Master's thesis). University of Nebraska - Lincoln.


A THESIS presented to the faculty of the Graduate College at the University of Nebraska in partial fulfillment of requirements for the degree of Master of Science, (Computer Science), under the supervision of Ashok Samal and James Schnable, Lincoln, Nebraska: December, 2016 .

Copyright (c) 2016 - Srinidhi Bashyam


Plant breeding and the development of new food production depend on accurate measurement of different phenotypes (observable physical traits) of a plant. The plant phenotypes play a very important role in the agronomic production. The successful computation of plant phenotypes largely depends on the determination of the architecture of the plant, i.e., the arrangement of its parts (leaves, stems, flowers, etc.) relative to each other, and how the size, shape, and positions of those parts change over time. Researchers and breeders extract valuable information from these types of data to make an informed decision on which individuals to advance to produce new, more productive crop varieties. The goal of this research is to develop fully automated software system to determine the plant architecture and subsequently compute phenotypes from the individual components of the plant based on analyzing plant image sequences.

The thesis introduces a novel algorithm to determine the plant architecture using a graph-based approach to detect, track and monitor the growth of leaves in a plant. Each tracked leaf has information of its position and its growth from emergence to the death of the leaf. This information is used to compute several novel phenotypes: stem angle, leaf length, leaf-node angle, mid-leaf curvature, apex curvature and leaf-integral area.

Preliminary results of the implemented algorithm for dynamic leaf tracking shows 92.31% tracking accuracy from day 1 to 25 on a set of maize plants that were imaged once daily. The accuracy of leaf length estimation was 94.4%. It would be straightforward to adapt this algorithm to large datasets of hundreds of plants such as those generated by modern automated phenotyping platforms to automatically compute the plant architecture and leaf-based component phenotypes.

Advisers: Ashok Samal and James Schnable