Progressive dynamic hyperspectral image classification method

The invention discloses a progressive dynamic hyperspectral image classification method. The progressive dynamic hyperspectral image classification method comprises the following steps of 1, reading hyperspectral image data; 2, calculating the difference between each pixel in the hyperspectral image and a local neighborhood pixel of each pixel, performing discontinuity detection by setting a proper threshold value, and dividing continuous points and discontinuous points; 3, by comparing the continuous points with the points of the known categories in the local neighborhood of the continuous points, cyclically classifying the non-classified continuous points; 4, cyclically classifying the discontinuous points and the continuous points which are not classified in the step 3; 5, classifying all the remaining to-be-classified points in the step 4; and 6, outputting a classification result. The invention provides the progressive dynamic hyperspectral classification method. Compared with most existing methods, the