Identification and mapping of weed density in a soybean field using DMS image

S. Gopalapillai, Lei Tian, L. Tang, C. E. Goering

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

DMS images of a soybean field acquired using an airborne color infrared (CIR) camera was processed and analyzed for deriving information on spatial weed density within a field. Image indices such as NDVI, SRI, and normalized NIR, red, and green were used to calibrate the image. The relative overlay position of image with respect to ground truth data was arbitrarily adjusted to minimize the image geo-referencing error. The overlay error was analyzed by correlating the weed data and image data at several resolutions ranging from 0.76 to 6.84 m/pixel. The errors associated with geo-referencing and GPS were very critical in high-resolution analysis of spatial data sets. The correlation between image indices and weed density changed considerably at different resolutions. The spatial pattern in the image data and weed data matched best at a resolution of 5.3 m/pixel. Weed density was modeled from image indices using artificial neural networks (ANN). This ANN model predicted weed density fairly accurately and resulted in an R 2 value of 0.87.

Original languageEnglish (US)
Title of host publication2000 ASAE Annual International Meeting, Technical Papers
Subtitle of host publicationEngineering Solutions for a New Century
Pages775-793
Number of pages19
Volume1
StatePublished - 2000
Event2000 ASAE Annual International Meeting, Technical Papers: Engineering Solutions for a New Century - Milwaukee, WI., United States
Duration: Jul 9 2000Jul 12 2000

Other

Other2000 ASAE Annual International Meeting, Technical Papers: Engineering Solutions for a New Century
CountryUnited States
CityMilwaukee, WI.
Period7/9/007/12/00

Keywords

  • ANN
  • CIR
  • High-resolution
  • NDVI
  • Weed density

ASJC Scopus subject areas

  • Engineering(all)

Fingerprint Dive into the research topics of 'Identification and mapping of weed density in a soybean field using DMS image'. Together they form a unique fingerprint.

Cite this