The utility of texture analysis to improve per‐pixel classification for high to very high spatial resolution imagery

Abstract Earth observation data are becoming available at increasingly finer resolutions. Sensors already in existence (IKONOS, Quickbird, SPOT 5, Orbview) or due to be launched in the near future will reach 1–5 m resolution. These very high resolution (VHR) data will provide more details of the urban areas, but it seems evident that they will create additional problems in terms of information extraction using automatic classification. In this framework, this paper examines the potential of the spectral/textural approach to improve the classification accuracy of intra‐urban land cover types. The utility of the textural analysis was measured in comparison with multi‐spectral per‐pixel classifications. Haralick's second‐order statistics were applied to the co‐occurrence matrix. Four texture indices with six window sizes created from panchromatic images were tested on images at high to very high resolutions (10–1 m). The results show that the optimal index improving the global classification accuracy is the homogeneity measure, with a 7×7 window size. Moreover, for 1 m images, texture measure of homogeneity allows one to decrease the shadows. Acknowledgements The authors wish to thank Prof. E. Lambin and Dr S. Henry, Remote Sensing & Land‐Use Changes Laboratory (Department of Geography), Université Catholique de Louvain, Belgium, for their technical support in this research.