In the present paper we review relationships between commonly used statistical approaches to analysis of image texture. The approaches considered characterize image texture by means of the statistics of grey- tone co- occurrence contrast, grey- tone co- occurrence correlation, semivariance, and autocorrelation. In the literature, the relationships between these textural measures are rarely discussed and sometimes overlooked. Grey-tone co- occurrence contrast and semivariance are shown to represent identical statistics. The expectations for the statistic of grey-tone co- occurrence correlation and the coefficient of autocorrelation are demonstrated to agree. Grey-tone variance is shown to link the statistics of grey-tone co-occurrence contrast and grey-tone co-occurrence correlation.
|Journal||Canadian Journal of Remote Sensing|
|Publication status||Published - 2005|