A FAST METHOD TO DIAGNOSING CITRUS GREENING FOR USE IN WARNING an‎d ALERT SYSTEMS

Citrus greening or Huanglongbing (HLB), become one of the most devastating disease in the worldwide citrus orchards. Multispectral imaging technique is used to classify diseased and healthy citrus trees in this study. Multispectral images were acquired by five discrete bands multispectral imaging camera embedded in an unmanned aerial system. Support vector machine was used to classify the samples using generated features including 10 vegetation indices and 5 bands. A total of 25 trees, 15 infected and 10 healthy trees were chosen by expert scouting as samples to train, validate and analysis of classifier. The overall classification results in the check samples was obtained 79.60% for SVM model. Therefore, it demonstrated that multispectral imagery has the great potential to be used for detection of HLB infected trees in citrus orchards