morphological changing information of volcanic ash cloud [10], [11].
Therefore, accurately remote sensing classification of volcanic ash
cloud has an important practical significance for improving the
disaster prevention and mitigation of volcanic ash cloud.
At present a variety of volcanic ash cloud detection from remote
sensing image have been presented with the new theory‘s in-depth
study, i.e., statistical theory, fuzzy theory and machine learning [12],
[13]. However, there is some common computation complexity,
application narrow and low accuracy in these mainstream
approaches. In contrast, the fuzzy C-means clustering (FCM) can get
each data sample‘s membership of all types of class centers and then
determine the ownership of sample points [14], [15]. This also has
great application potential in the volcanic ash cloud detection. As a
typical dynamic clustering algorithm, in essence FCM seeks an
optimal solution by the gradient descent method. However, the
objective function which corresponds to the results often may be a