Abstract: Aiming at the problem of how to extract the plant disease area accurately, a method based on EM and K-means hybrid clustering is proposed. The method can effectively extract the leaf target and have some screening effect on the more complicated background, which is superior to other classical methods in the case of obvious difference between the target and the background. Based on the chlorotic characteristics of the plant disease area, the K-means method was used to combine the Lab color space. Using the uniformity of the color distribution of the Lab color space, the A component was extracted as the reference component, and the disease area was extracted from the leaf target. The experimental results showed that the extraction of plant diseases based on EM and K-means hybrid clustering method is feasible.

Key words: plant disease area, EM algorithm, Lab color space, K-means algorithm, hybrid clustering