Development of urban forest monitoring method and land cover monitoring in Takamatsu

Atsuko Nonomura(1) and Takuro Masuda(1)

(1) Kagawa University, Hayashicho 2217-20, Takamatsu, 761-0396, Japan


The traditional image classification techniques, which are based on spectral information of each pixel, are normally considered ineffective in urban land-use and land cover features in high resolution images due to the variation in the spectral response within one category. The spectral characteristic of each category may exhibit very high standard deviation and it might be difficult to classify whole data into land cover categories. In order to identify the spectrally heterogeneity of each land cover type, quantifying the spatial pattern of spectral characteristics is indispensable. In this study, we explored the method to quantify the vegetation distribution in urban area by using both texture and spectral information. For examining spatial pattern, several methods have been used to quantify the spatial distribution; fractal geometry, entropy, lacunarity, etc. Although fractal dimension has been commonly used to quantify the pattern (Alados et al., 2005), sometimes it cannot identify the difference of even greatly different appearance pattern (Plotnick et al., 1993). On the other hand, lacunarity can quantify the geometric arrangement of gaps (McIntyre and Wiens, 2000). Therefore, in this study, lacunarity was calculated using ALOS/PRISM image and it was merged with spectral information computed with ALOS/AVNIR-2 image.


Symposium presentation


  Higher level                 Last modified: 07.05.06