Comparison between three methods used in automation of the steps of arbuscular mycorrhizal fungi spore count process
The most varied species of organisms in the environment establish forms of association, the main purpose of which is survival. The mutual association between arbuscular mycorrhizal fungi (AMF) and most plant roots is called arbuscular mycorrhizal (AM). Using these associations and with the previous knowledge of the quantity of AMF spores, it is possible to define the appropriate quantities for the application of fertilizers in a given area, reducing their waste. The aim of this study is to propose and to investigate the efficiency of a semi-automated count method, using artificial neural network (ANN) and convolutional neural network (CNN)-based models, as identifiers and classifiers of AMF spore images. The study evaluates the performance of this method. The circle Hough transform (CHT) is used as a processing tool for the images to be classified by ANNs or CNNs. The results show that the developed models are viable classifiers, with classification rates of around 100%, when compared to the manual count method.
Index Terms– pattern classification, semi-automated counting, artificial neural network, image processing, circle Hough transform.
