Statistical and Neural Network Classifiers for Citrus Disease Detection Using Machine Vision

Abstract

The citrus industry is an important constituent of Florida's overall agricultural economy. Proper disease control measures must be undertaken in citrus groves to minimize losses. Technological strategies using machine vision and artificial intelligence are being investigated to achieve intelligent farming, including early detection of diseases in groves, selective fungicide application, etc. This research used a texture analysis method termed the color co-occurrence method (CCM) to determine whether classification algorithms could be used to identify diseased and normal citrus leaves. Normal and diseased citrus leaf samples with greasy spot, melanose, and scab were collected in the field and brought to the laboratory for the development of suitable segmentation and classification algorithms. Four feature models were created for classification analysis using varying subsets of a 39-variable texture feature set. The classification strategies used were based on a Mahalanobis minimum distance classifier, using the nearest neighbor principle, as well as neural network classifiers based on the back-propagation algorithm and radial basis functions. The leaf sample discriminant analysis using the Mahalanobis statistical classifier and the CCM textural analysis achieved classification accuracies of over 95% for all classes (99% mean accuracy) when using hue and saturation texture features. Likewise, a back-propagation neural network algorithm achieved accuracies of over 90% for all classes (95% mean accuracy) when using hue and saturation features. It was concluded that the Mahalanobis statistical classifier and the back-propagation neural network classifier performed equally well when using ten hue and saturation texture features selected through a stepwise variable reduction method. Future studies will seek to apply the developed algorithms in a natural citrus grove environment.

Multiple Classifier Combination For Recognition Of Wheat Leaf Diseases

“…Ahmad [3] extracted features from different color space: RGB, HIS and rgb (normalized RGB) to evaluate the color change of maize which lacks water or Nitrogen. Pydipati [13] chose HIS color model and applied chrome Co-occurrence Matrix as texture features to recognition of citrus leaf diseases. So firstly regions of infected spots were extracted from the leaf, and then used information of image to extract the base-level features such as color features, texture features, shape features.…”

“…However, few researchers in the field of crop diseases diagnosing consider these techniques for pattern recognition. Most of them used single classifier or learning techniques such as SVM [11], Bayesian classifier [12], Neural Network [13], etc.…”

Classification of Rubber Tree Leaf Disease Using Spectrometer

“…The optimized ANN model was then evaluated and validated through analysis of the performance indicators. Findings in this work have shown the both models have produced about 70% in diagnostic accuracy with more than 80% achievement for sensitivity [3] [4].…”

Comparison of Image Texture Based Supervised Learning Classifiers for Strawberry Powdery Mildew Detection

“…Wang et al [32] reported improved control of tomato diseases by predicting late blight infections using ANNs. Pydipati et al [33] utilized backpropagation ANNs algorithm and colour co-occurrence textural analysis for citrus disease detection and achieved classification accuracies of over 95% for all classes. The researchers also claimed that an overall 99% mean accuracy was achieved when using hue and saturation textural features.…”