Severity Estimation of Plant Leaf Diseases Using Segmentation Method

Plants have assumed a significant role in the history of humankind, for the most part as a source of nourishment
for human and animals. However, plants typically powerless to different sort of diseases such as leaf blight, gray
spot and rust. It will cause a great loss to farmers and ranchers. Therefore, an appropriate method to estimate
the severity of diseases in plant leaf is needed to overcome the problem. This paper presents the fusions of the
Fuzzy C-Means segmentation method with four different colour spaces namely RGB, HSV, L*a*b and YCbCr
to estimate plant leaf disease severity. The percentage of performance of proposed algorithms are recorded and
compared with the previous method which are K-Means and Otsu’s thresholding. The best severity estimation
algorithm and colour space used to estimate the diseases severity of plant leaf is the combination of Fuzzy
C-Means and YCbCr color space. The average performance of Fuzzy C-Means is 91.08% while the average
performance of YCbCr is 83.74%. Combination of Fuzzy C-Means and YCbCr produce 96.81% accuracy. This
algorithm is more effective than other algorithms in terms of not only better segmentation performance but also
low time complexity that is 34.75s in average with 0.2697s standard deviation.