Automatic detection of rust disease of Lentil by machine learning system using microscopic images

by: International Journal of Electrical and Computer Engineering (IJECE)

Publication date: 2019-02-01

Usage: Attribution-ShareAlike 4.0 International

Topics: Brightness, Bi-Histogram, Equalization, Lentil, Local binary pattern, Microscopic image, Rust disease

Collection: journals_contributions; journals

Accurate and early detection of plant diseases will facilitate mitigate the
worldwide losses experienced by the agriculture area. MATLAB image
processing provides quick and non-destructive means of rust disease
detection. In this paper, microscopic image data of rust disease of Lentil was
combined with image processing with depth information and developed a
machine learning system to detect rust disease at early stage infected with
fungus Uromyces fabae (Pers) de Bary. A novel feature set was extracted
from the image data using local binary pattern (LBP) and HBBP (Brightness
Bi-Histogram Equalization) for image enhancement. It was observed that by
combining these, the accuracy of detection of the diseased plants at
microscopic level was significantly improved. In addition, we showed that
our novel feature set was capable of identifying rust disease at haustorium
stage without spreading of disease.