26 July 2007 Classification of Landsat 7 ETM+ imagery in western mountainous area of Zhejiang based on gray-gradient co-concurrency matrix
Author Affiliations +
Texture feature is becoming more and more important for classification of remote sensing image, especially in remote sensing image in mountainous area. An approach to classification of western mountainous area of Zhejiang land cover using ETM+ imagery is described in this paper. Firstly the gradient images of research area were obtained using different edge detection methods with Roberts, Sobel, Prewitt and Canny operator using ETM+ pan image. The results of four different edge detection methods were evaluated qualitatively and quantitatively. The qualitative evaluations mainly considered the visual effect so that the results of combining edge images with original image for qualitative evaluation, and the edge points,4-connected component quantities and 8-connected component quantities were adapted to quantitatively evaluate different edges. Then Canny operator was selected as the gradient operator according to the qualitative and quantitative evaluation results of different research area's edge images and fifteen texture features were obtained based on the Gray-Gradient Co-concurrency Matrix consequently through MATLAB programmer with the Canny operator. Finally, the classification results based on the spectral respond feature only and the texture feature with the spectral respond feature were evaluated separately. It shows that the texture features highlight the residents, rivers etc. which the geometric structure of space themselves are more obvious than others; enhancing the undulating the distinction between water and the shadow.
© (2007) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Feng Li, Feng Li, Manchun Li, Manchun Li, Yongxue Liu, Yongxue Liu, } "Classification of Landsat 7 ETM+ imagery in western mountainous area of Zhejiang based on gray-gradient co-concurrency matrix", Proc. SPIE 6752, Geoinformatics 2007: Remotely Sensed Data and Information, 67523M (26 July 2007); doi: 10.1117/12.761253; https://doi.org/10.1117/12.761253

Back to Top