Figure 7From: Classification of hyperspectral imagery with neural networks: comparison to conventional toolsTypical misclassifications. Examples of typical misclassifications by the MLH classifier, from left to right, for a class J, a class Q, and another class J sample. In each panel, the top plots show how the correctly classified sample pattern (thick line) fits into the envelope of the training samples (thin dotted lines) of the corresponding class. The bottom plot shows how the same sample fits with the envelope of the class into which MLH misclassified the pattern.Back to article page