C. ACCURACY ASSESSMENT
In order to assess the volcanic ash cloud remote sensing classification with the proposed method, in the experiment, 500 data points were initially randomly generated by our program. Next, the accuracy assessment of the classified volcanic ash cloud is performed with confusion matrix. Two kinds of evaluation metrics are applied in our experiment, including Kappa coefficient and overall accuracy (see Table 4 and 5).
As one of the most important contents of volcanic ash cloud monitoring and aviation safety application, at present there are many sophisticated volcanic ash cloud classification methods, such as SWTD and RGB PC method. Similarly, in this study the SWTD method and RGB PC method are used to classify the same Sangeang Api volcanic ash cloud images on 30 May 2014 so as to assess the