Analysis of classification accuracy for pre-filtered multichannel remote sensing data

Highlights：

• We propose to classify multichannel remote sensing images using pre-filtering.

• DCT-based block filtering is used to suppress signal dependent noise in images.

• Radial basis function neural network and support vector machines are employed.

• Different cases of learning are considered: using noise-free, noisy and pre-filtered image.

• The use of the pre-filtered image for training produces better classification results.