skip to main content


Title: Point Cloud Based Approach to Stem Width Extraction of Sorghum

A revolution in the field of genomics has produced vast amounts of data and furthered our understanding of the genotypephenotype map, but is currently constrained by manually intensive or limited phenotype data collection. We propose an algorithm to estimate stem width, a key characteristic used for biomass potential evaluation, from 3D point cloud data collected by a robot equipped with a depth sensor in a single pass in a standard field. The algorithm applies a two step alignment to register point clouds in different frames, a Frangi filter to identify stemlike objects in the point cloud and an orientation based filter to segment out and refine individual stems for width estimation. Individually, detected stems which are split due to occlusions are merged and then registered with previously found stems in previous camera frames in order to track temporally. We then refine the estimates to produce an accurate histogram of width estimates per plot. Since the plants in each plot are genetically identical, distributions of the stem width per plot can be useful in identifying genetically superior sorghum for biofuels.
 [1] ;  [1]
  1. Univ. of California, Berkeley, CA (United States)
Publication Date:
Grant/Contract Number:
Accepted Manuscript
Journal Name:
Electronic Imaging
Additional Journal Information:
Journal Volume: 2017; Journal Issue: 17; Journal ID: ISSN 2470-1173
Research Org:
University of Illinois at Urbana-Champaign, IL (United States)
Sponsoring Org:
USDOE Advanced Research Projects Agency - Energy (ARPA-E)
Country of Publication:
United States
97 MATHEMATICS AND COMPUTING; 09 BIOMASS FUELS; Frangi Filter; Genomics; Genotype-Phenotype Map; Phenotype Data Collection
OSTI Identifier: