STILL VIDEO IMAGE-ANALYSIS OF CROP RESIDUE SOIL COVERS

A still video system was used to take RGB color images for system calibration, procedural developments, and initial analyses of percentage of soil covered by crop residues. A state-of-the-art PC image processor was used with customized software. A simplistic discrimination technique was used to classify individual image pixels as either residue cover or soil background. It was determined that analysis of only 12% of the image pixels would provide stable results. Shadows were not a significant problem. Color of the background affected the results, even with residues simulated with wooden dowels of known dimensions. Relative size of the residue pieces affected the performance, with better performance achieved with thick residues, such as corn, as contrasted with wheat residues. Analyses of actual residue images were generally too variable to be acceptable without further development on the discrimination of colors and soil backgrounds.