Texture feature extraction of mountain economic forest using high spatial resolution remote sensing images

The inventory of economic forest planting on mountainous area is of great interest for the shareholders, ecologists, and governors. This study presents a novel object-based remote sensing image texture extraction method to aid the classification of mountain economic forest. Whereas the texture pattern of man-planted forest on mountainous area are similar with human fingerprint on remote sensing images, then we integrated the fingerprint recognition method with the image object-based GLCM (gray-level co-occurrence matrix) texture extraction method to classify economic forest plantation. The given method firstly enhances the texture feature of segmented image objects using a 2D Gabor filter; and then carries out an image binaryzation process to the filtered image objects; it lastly uses GLCM to calculate the texture characteristics of each image object. The classification result of the presented method shows a considerable classification accuracy increment comparing to the image object-based GLCM texture extraction method.