ABSTRACT
This chapter explores the basic image processing techniques applicable to plant phenotyping analysis. It includes techniques to isolate (segment) a plant from its background, to recover the structure of the plant, and to derive properties from the segmented plant image. The chapter discusses processes to compute the phenotypes of two common architectures found in model plants, namely the rosette-like architecture in Arabidopsis or tobacco, and the distichous phyllotactic architecture in maize or sorghum. It describes algorithms to compute phenotypes for most other plants that can be adapted using image processing operations. The computation of plant phenotypes, using computer vision-based techniques, enables automatic extraction of plant traits from the images. The input to the processing pipeline is typically an image of the plant. The input image may contain other objects in the scene, e.g., the plant container, or exposed soil.
