Processing remote sensing imagery for accurate and complete information extraction requires specialized software and methodologies; GISmatters has decades of experience in this field, using off-the-shelf tools and developing custom image-processing software with a wide range of data and formats, from black-and-white aerial photos to hyperspectral imagery - even medical imagery! We can take the data from raw digital values to fully rectified, georegistered, and processed information products.
 
 

Remote Sensing: Transformations

Principal components analysis of image
Roll over images to compare views
Click to see larger in another window
Image data is often transformed in either the spectral or spatial domains for purposes of information extraction, image compression, or image filtering. Operations such as smoothing, edge detection, or radiometric or geometric enhancement may not be possible in the original image but are straightforward on the transformed data. In the spatial domain, common methods include Fourier and wavelet transforms; in the spectral domain they include Principal Components Analysis (also known as Empirical Orthogonal Functions), tasseled cap, and various band ratios like NDVI.

The images at right show a natural-color (when you roll the mouse pointer over the image) and Principal Components version of a Landsat scene. The PCA version displays the first 3 principal components with red, blue, and green colors, respectively. This transform maximizes variance in the first components and eliminates correlation between the components, making it well suited for information- rich image display, image classification, and image compression.