Influence of weed maturity levels on species classification using machine vision

Abstract

The environmental effect of weed control systems has stimulated research into new practices for weed control, such as selective herbicide application methods on weed-infested crop areas. This research used the color co-occurrence method (CCM) texture analysis to determine the effects of plant maturity on the accuracy of weed species classification of digitized images. Two different experimental combinations of weed species and maturity level were examined. The weed species evaluated were ivyleaf morningglory, giant foxtail, large crabgrass, and velvetleaf, with soil image sets added to each experiment. One study examined classification accuracies for two weed species at three maturity levels, and the second study examined four weed species at two maturity levels. For each species-maturity level combination, 40 digital images were collected from a manually seeded outdoor plant bed. Digitized images were transformed from the red–green–blue (RGB) color format into hue–saturation–intensity (HSI) format to generate CCM texture feature data. Stepwise variable reduction procedures were used to select texture variables with the greatest discriminant capacity. Then discriminant analysis was used to determine the classification accuracy for the two different experiments. When using HSI texture statistics, discriminant analysis correctly classified weed species within and across maturity levels with an accuracy above 97% for both experimental groups. These image processing algorithms demonstrate potential use for weed scouting, weed infestation mapping, and weed control applications using site-specific farming technology.

Potential use of ground-based sensor technologies for weed detection

“…A classifier was trained with texture features generated from theses CCM matrices. Burks et al achieved 100% accuracy in soil classification, while the classification accuracy between the five other plant species exceeded 90% . Using the texture of weed plants similar results were achieved and in other weed detection cases …”

“…Changes of filters led to recent developments of NDVI cameras based on standard RGB camera technology . Many researchers have applied machine vision systems for the detection of weeds in agricultural fields . Some of the developed systems cannot just identify weed patches based on the fact that weed‐infested spots contain more biomass than non‐infected ones, but can also differentiate weeds from crops and among different weed species with varying degrees of success .…”

“…With a colour co‐occurrence method (CCM), the texture of weed plants was used to differentiate species . The colour space was transformed from RGB (red, green, blue) values to HSI (hue, saturation and intensity) values.…”

Detection of Italian Ryegrass in Wheat and Prediction of Competitive Interactions Using Remote-Sensing and Machine-Learning Techniques

“…Given the pale green color of ryegrass leaves compared to that of wheat, the difference in the greenness level was obvious with hue and saturation values. Several studies have credited hue and saturation for their ability to differentiate plants based on the greenness level [44,45]. Additionally, ExG was shown to be useful in separating plant tissues from other backgrounds (soil and weathered plant residue) [46].…”

The Biological Engineer: Sensing the Difference Between Crops and Weeds

“…The name "co-occurrence matrix" is used for the resulting matrix of conditional probabilities created for all combinations of each pair of co-occurring grayscale or color values at the defi ned spatial offset and direction. In the majority of the published studies investigating the use of texture for weed detection, the images contained a monoculture (i.e., a single species per image) and classifi cation rates of 90 % or more were frequently obtained (Shearer and Holmes 1990 ;Meyer et al 1998 ;Burks et al 2002 ;Tang et al 2003 ;Ishak et al 2009 ).…”