Abstract:
Turcicum is caused by the fungus (Exserohilum turcicum (Pass.) Leonard & Suggs) and
appear on the leaves of maize at any stage of growth. The turcicum leaf blight injures or kills
the maize leaf tissues and thereby reduces the area of green chlorophyll which
manufactures/synthesise food for the plant. Traditionally, application of fungicides is carried
out manually without objectively quantifying the severity of the disease caused by turcicum
leaf blight leaf spot. In this research work, digital image processing techniques was used to
estimate the disease severity of live-maize leaves by extracting the area feature. Total area of
leaves and area of diseased region of leaves were extracted from 264 maize-leaf images
captured in four rounds. First round was before spraying fungicide, second round was one
week after the first round fungicide sprayed, third round was one week after the second time
fungicide sprayed and forth round was a week after the third time fungicide sprayed. Control
group was the one on which fungicide was not sprayed (group one). Then, the RGB images
were converted to CIELAB color space and the detection of turcicum leaf blight was
performed using the k-means clustering to estimate the disease severity of the leaves. MatLab
based algorithms were developed to determine the total area and infected lesion area of the
leaf samples. The severity levels leaves were also assessed by two experts, visually. In most
samples, the two experts scored different values for the same sample leaves. The optimum
volume of fungicide is obtained at group 7 and 8 which means the plot of maizes on which
70.89ml, and 74.25ml fungicide. The accuracy of the algorithm developed for estimation of
disease severity using image technology was 95.6%. Thus, atomated measurement of disease
severity on maize leaf gives accurate result. The result of algorithms developed shows that the
possibility of measuring Turcicum leaf blight of maize leaf and estimating its optimum volume
of fungicide.
