Other Free Encyclopedias » Online Encyclopedia » Encyclopedia - Featured Articles » Contributed Topics from K-O

Multispectral Imaging

data channels spectral images

DAVID MALIN
Anglo-Australian Observatory, RMIT University

Multispectral imaging involves making images using more than one spectral component of energy from the same region of an object and at the same scale. It is clearly not a new method of imaging; color film is a three-channel multispectral detector and special infrared-sensitive color films have long been available to achieve the above-mentioned condition. A contemporary definition of multispectral imaging implies that the channels or passbands will be separate, and that they may contain image data obtained outside the range of sensitivity of photographic materials. Often there may be more than three sets of data. Multispectral imaging may also be quantitative, in that the separate channels are well-defined spectrally and often calibrated radiometrically. They may also have a temporal element, in that the images may be made at different times.

In practice, each channel of a multispectral image may be displayed as a grayscale representation, or in combinations of two or three channels as a color composite image. If the data channels contain images made in blue, green, and red light, a true-color picture can be made; however, if any or all of the channels contain data from outside the visible spectrum, the result will be a false-color image. Separate channels may also be subtracted from each other to reveal subtle differences between them. Much of this was not practicable before the digital imaging revolution, and reflects the dual nature of multispectral imaging, as a source of spectral data to be analyzed and of images to be visualized. This in turn determines how the data are handled.

The information obtained from multispectral images can be from self-luminous sources such as the sun, stars, and nebulae or reflective sources such as the planets (including the earth), documents, biomedical subjects, or foliage. The data may be both radiometric, recording brightness or intensity in a broad, defined passband, or may consist of many narrow passbands, recording the spectral energy distribution at higher resolution. This is in addition to the texture, geometry, and context that would normally be expected from images.

Interpretation of a multispectral image requires an understanding of the characteristics of the filter-detector combination used to obtain the data so that the spectral signature of the scene can be recovered. The multispectral imaging systems themselves operate over a broad range of wavelengths, from ultraviolet to the visible and into the thermal infrared (200 to 15,000nm), so the types of detectors used vary widely.

Development of multispectral imaging includes superspec-tral imaging, which can involve ten or more spectral channels, each with narrow bandwidths, enabling greater spectral resolution. Developments also include hyperspectral imaging or imaging spectroscopy, which can make images in a hundred or more contiguous spectral bands. However, as the radiation collected from the scene is divided into ever narrower channels both the spatial resolution and signal-to-noise may be degraded.

The main application of these multispectral systems is in remote sensing, especially from earth-orbiting satellites and aircraft. Multispectral imaging is especially useful in astronomy, agriculture, oceanography, geology, pollution monitoring, and mine-laying detection, but also in heat sensing, medicine, and forensic work, including microscopy, and the copying of documents for archival purposes.

Murchison, Sir Roderick Impey [next] [back] Multiplication of the Loaves and Fishes

User Comments

Your email address will be altered so spam harvesting bots can't read it easily.
Hide my email completely instead?

Cancel or