Research on Airborne High Resolution SAR Image Classification.

This paper studies the gray feature and texture feature including initial moment, energy based on gray level co-occurrence. An approach is proposed that feature is extracted and selected. Furthermore the BP neural network is applied to the image supervised classification. At least, the small areas are removed by morphological open operator. Considering the gray feature and texture feature of the SAR image, the method is more suitable for SAR image classification than the traditional method, which uses the texture feature only. The experimental results show the method can solve the airborne high resolution SAR image classification perfectly.