Introduction to statistical pattern recognition (2nd ed.)
Introduction to statistical pattern recognition (2nd ed.)
Hi-index | 0.00 |
When principal axis transformations are applied to multispectral scanner (MSS) data, the majority of data variability is shown to be contained in the first two or three components. This paper describes a method for generating a color composite picture from these components whereby most of the information collected by an MSS can be conveyed in a single color picture. The first component is found to be a weighted sum of image data from all channels, and therefore it is natural to associate the first component with brightness. To be consistent with this interpretation, only the first component of the transformed data is used to control the brightness of the color composite picture.