Most remote sensing data exist as digital images, and appropriate image processing allows the emphasis of certain aspect and subsequent extraction of information for specific applications. A digital image is a representation of the reality as a grid of picture elements. It can be considered as an array of numbers that can be stored and handled by a digital computer. The picture elements are pixels and each pixel has a specific value (usually in grayscale). This value is a digital number (DN), which usually represents the amount of energy recorded by the sensor at this pixel position or any other characteristic recorded by the sensor, e.g. elevation. Each row of the image grid, or matrix, corresponds to one scan line. Each pixel is characterized by its row r and column c position in the image, as well as by its value. Additional geographical information is needed in order to assign a geographic location to a pixel. The digital number are integers usually compressed in one byte (= 8 bit) representation, i.e. each pixel can take 256 values. Digital images are raster data, as opposite to vector data. Whereas vector data can be points, lines or polygones, raster data always consist of pixels. A pixel is the smallest element in which an image can be divided into. The pixel size varies depending of the instrument and of the sampling used. Large pixel may contain information about several objects of the recorded scene. However, they only have one value. These are called mixed-pixel, as e.g. several land cover classes are represented within one pixel and they cannot be distinguished from another. In multispectral imagery, each region of the electromagnetic spectrum is recorded in an independent image (band). Therefore, at a specific array position (r,c), there exist several pixels, each with a specific value corresponding to the energy recorded for the considered band. This result in a three-dimensional matrix. The bands of a multispectral image can be displayed three at a time in the computer using for each band one of the three primary colors red, green and blue (RGB). This is called a color composite image. If the color composite represents a combination of the visible red, green and blue bands in their respective color, the combination is called natural or true color composite, as it corresponds to what the human eye sees naturally. Any other combination, for example considering bands of wavelengths that are not visible for the human eye is called a false color composite. It is often used to highlight the spectral differences and particular image features in order to extract information.
Discuss main characteristics of digital imagery
Completed