Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

The possible spread of late blight from volunteer potato plants requires that these plants being removed from fields. However, because of high labour, energy and chemical inputs associated with this removal process, an automatic detection and removal system becomes necessary. In this paper, the development and comparison of two colour-based machine vision algorithms for in-field volunteer potato plants detection in sugar beet fields were reported. When classification accuracy was evaluated at plant level, an Adaptive Neural Network classifier and a joint classifier of K-means clustering and Bayes classification produced closely matched results. Specifically, from 192 top view images, 92% of volunteer potato plants were correctly detected both methods. There were 4% sugar beet plants being wrongly identified as volunteer potato plants, which was largely caused by occlusions of leaves. At pixel level, K-means/Bayes classifier gave slightly better results on both top view and slant view images. Although K-means/Bayes with a static lookup table gave slightly better results, an adaptive neural network could be more suitable for the changing conditions in the fields. Especially for the case of using an outdoor autonomous robot for volunteer plants removal, adaptive methods possesses a greater potential.