Extraction of golf course based on texture feature of SPOT5 image

Based on SPOT5 remote sensing image in 2008, we selected two golf courses-one in urban area of Shenzhen and the other in the forest area of Shenzhen, as the two study areas. First, we made principal components analysis of the two areas for data compression and enhancing geometric information. Second, we processed the image and filtered the noise by the wavelet transformation, and the textures of the SPOT5 images were analyzed using Gray Level Co-occurrence Matrices. Based on these analyses, we selected four statistic indexes (contrast Contrast, homogeneity, correlation and entropy.) Finally, with the selection by man-machine interpretation, we chose the optimal threshold for image segmentation and information extraction of the golf course. The results showed that the method of texture feature extraction is better in classifying the land types during information extraction of the golf course and providing more precise results compared to the traditional method.