How are multispectral image bands often highly correlated?
Robert A. Schowengerdt, in Remote Sensing (Third edition), 2007 Multispectral image bands are often highly correlated, i.e., they are visually and numerically similar (Fig. 5-7 ). The correlation between spectral bands arises from a combination of factors: FIGURE 5-7. Two-band scattergrams of different images from different sensors.
How many spectral bands are there in an image?
It is also possible to capture hundreds of wavelength bands for each pixel in an image. Multispectral imaging captures a small number of spectral bands, typically three to fifteen, through the use of varying filters and illumination. Many off-the-shelf RGB cameras will detect a small amount of Near-Infrared (NIR) light.
How are multispectral image data used in classification?
The multispectral image data can be used directly in a classification, but it contains all the various external influences, such as atmospheric scattering and topographic relief, described in earlier chapters. Also, the data are often highly correlated between spectral bands, resulting in inefficient analysis.
Where is the most contrast in a multispectral image?
For the nonvegetated scene, the major spectral contrast appears to be in bands 5 and 7 (where mineral signatures are found; see Fig. 5-6 ), while for the vegetated scene, bands 3, 4, and 5 contain most of the spectral contrast because of the vegetation reflectance “edge” at 700nm.
How is a multispectral image optimally redistributed?
It optimally redistributes the total image variance in the transformed data. The first PC image contains the maximum possible variance for any linear combination of the original bands, the second PC image contains the maximum possible variance for any axis orthogonal to the first PC, and so forth.
When was the multispectral image ( MNF ) transform introduced?
Also known as the Noise-Adjusted Principal Components (NAPC) transform (Lee et al., 1990), this technique was introduced in Green et al. (1988). The MNF is a modification of the conventional PCT to improve the isolation of image noise that may occur in one or only a few of the original bands.