Penggunaan Vis-NIR untuk Deteksi Serangan Huanglongbing pada Daun Jeruk

Raden Arief Firmansyah • Kudang Boro Seminar • Widodo Widodo
Journal article Jurnal Keteknikan Pertanian • 2017


Huanglongbing is citrus disease which is a major threat for citrus orchard. Neither disease has a cure nor an efficient means of control. Early detection is important to prevent development and spread of the disease. The most effective detection used DNA test by PCR. However, identification used DNA test required sample preparation, time-consuming and expensive. The objective of this study was to build detection of healthy and HLB-infected leaves software. The leaf samples collected from citrus orchard in Situgede village, Bogor. Sampleleaves divided into three group, Huanglongbing-infected leaves, healthy leaves and asymptomatic leaves. All samples was tested by PCR for verification visual symptoms of huanglongbing. Vis-NIR spectrometer with a spectra range of 339 to 1022nm was used to acquisition HLB-infected and healthy leaves spectral data. MSC, SNV, baseline correction, first and second derivative were used for pretreatment method. Artificial neural network was used to build classification model. X-loading plot from principal component analysis was used to obtain sensitive wavelength. Classification for healthy and HLB-infected classs used sensitive wavelength baseline correction-based had the best performance and high accuracy (100%). The classification model was embedded in software PC-desktop based which was used visual basic programming language. Asymptomatic leaves spectral from HLB-positive tree were used to testing classification model. Model classified data into HLB-infected group, which was consistent with PCR test. The result from this study indicated that developed software could be used to HLB detection in early stage of disease.


  • 275 kali dilihat
  • 121 kali diunduh


Jurnal Keteknikan Pertanian

Jurnal Keteknikan Pertanian pada awalnya bernama Buletin Keteknikan Pertanian, merupakan publikas... tampilkan semua