Normalisation of multi-date digital remote sensing data using scene statistics

Some of the studies involving digital temporal data require normalisation. Many of the normalisation methods need information on prevailing atmospheric conditions, sensor-related details etc., whereas some other methods use only the information which are within the data set. A study was conducted in Kudremukh Iron Ore miningsite, Karnataka using multi-date data from Landsat MSS of 1976 and 1985. A normalisation method using only the scene statistics of the images is suggested which brings the multi-date data at par. This method is compared with some of the other methods which use only the information content of the data set. Comparison of these methods was made by scattergrams, distance measure etc. By generating scattergrams for various combinations of non-normalised and normalised images, different methods were compared. Based on ground information, some changed and unchanged classes were noted on the various normalised and original images. Using the distance measure for classes in different images, the normalisation procedures were compared. The proposed method turned out to be the best among the methods compared.