Autocorrelation of Control Points on 11-Band Multispectral Imagery

The use of multispectral imagery with other types of imagery requires that the combined set be accurately registered. This study investigates correlation properties between bands of an 11--channel multispectral scanner for 37 control points in support of interband registration that can later be extended to dissimilar image registration. The results show a high degree of correlation between channels 1 through 7, but a lower and unpredictable correlation for other combinations. In addition to providing information about interband registration, the results also have implications for multispectral segmentation algorithms. Keywords: Autocorrelation measures, Image registration, Multispectral imagery, Feature extraction, Image segmentation algorithms.