Skip to main content
ARS Home » Midwest Area » Columbia, Missouri » Plant Genetics Research » Research » Publications at this Location » Publication #197319
Title: A COMPUTATIONAL APPROACH FOR CHARACTERIZING STANDARDIZED PHENOTYPIC IMAGES FOR MAIZE

Author
item SHYU, CHI-REN - UNIVERSITY OF MISSOURI
item GREEN, JASON - UNIVERSITY OF MISSOURI
item FARMER, CYNTHIA - UNIVERSITY OF MISSOURI
item KAZIC, TONI - UNIVERSITY OF MISSOURI
item COE JR, EDWARD - USDA RETIRED
item Schaeffer, Mary
item Millard, Mark
item Cyr, Pete
item Gardner, Candice

Submitted to: Maize Genetics Conference Abstracts
Publication Type: Abstract Only
Publication Acceptance Date: 2/25/2006
Publication Date: 3/9/2006
Citation: Shyu, C., Green, J., Farmer, C., Kazic, T., Coe Jr, E.H., Schaeffer, M.L., Millard, M.J., Cyr, P.D., Gardner, C.A. 2006. A computational approach for characterizing standardized phenotypic images for maize [abstract]. 48th Annual Maize Genetics Conference. 48:56.

Interpretive Summary:

Technical Abstract: Experimentation with mutant maize plants is an effective method for understanding the roles of specific genes as well as for visualizing the phenotypic effects of these mutations. For visually observed phenotypic effects, annotations are made by scientists to document the physical state of the mutated plant; however, the language used to describe the mutations can be vague, especially in terms of color, texture, and size (e.g., the leaf is pale green, the kernel is variegated, the plant is short). Color descriptions are further complicated by the fact that "light green" to one person may be described as "yellow green" by another. To combat this vagueness or uncertainty in mutant descriptions, image processing and computer vision algorithms can be developed to quantify these types of visual features, eliminating the subjective component of human perception in these kinds of descriptions. We are developing a web-based phenotypic information management system, VPhenoDBS, that will use these features to allow biologists to perform complex queries (query by image example, query by text annotation/ontology, and query by physical and genetic map information) on maize images. The web-based system will be publicly accessible to the plant community, particularly for the Maize community for the initial stage. We propose simple standards to capture phenotypic images for various body parts and development stages using commercially available digital cameras, color palettes, rulers, and homogeneous background settings under a consistent lighting condition. All images deposited to the VPhenoDBS using the simple standards, along with their corresponding text annotations, will be searchable and cross referenced to various maps with a unique visualization tool.