Abstract:Texture analysis has become an important means for improving the accuracy of remote sensing image
classification. As the texture feature is closely related to image scale, the determination of a scale for texture
analysis applied in remote sensing image classification is very important and corresponds to the choice of an
appropriate size of texture window for gray co-occurrence matrix texture analysis. The authors studied the spatial
relationship between the adjacent pixels in the remote sensing image, and selected the lag distance of the semi-
variogram that was determined when the value of the semi-variogram tended to be constant as the co-occurrence
window size. Under the restraint of the Maximum Likelihood supervised classification results, the co-occurrence
features were computed with a timely changeable co-occurrence window size according to the semi-variogram
analysis. This paper introduced a method of reasonable scale texture analysis for remote sensing image
classification and had an image taken in Changping District, Beijing as an example. The texture feature was
extracted from SPOT5 remote sensing data in the Titan Image secondary development environment and involved in
classification. A comparison of the results using the method proposed in this paper shows that the classification
accuracy has been improved effectively.