Abstract :
A computerized system for processing ERTS images for geologic applications is described. The basic philosophy of the system is to generate the highest possible quality images and to apply simple enhancement techniques optimizing images for direct human interpretation. Automated mapping, classification and analysis techniques are avoided for several reasons. First, several types of noise are present: system noise generated by the camera system, data transmission and data processing and "noise" inherent to the natural scene-cultural effects, vegetation masks and clouds and water. Second, the geologic environment itself is extremely complex. Automated interpretive techniques usually merge such noise undecipherably into the image. A major part of the processing described here is devoted to the removal of "noise" patterns, atmospheric effects, variable illumination, geometric corrections, and correction for high frequency loss. The enhancement stage is the one tailored for the specific application and scale; utilizes relatively simple algorithms; and consumes a lesser portion of the processing time. It has been found, further, that the image processing technology is ideal for handling other kinds of mappable data, such as gravity and magnetics, in which extremely large (107 - 109 bits) data sets can be treated displayed and analyzed as images. Examples include shaded relief models of "surfaces" representing geophysical data, color composites of multivariate data to examine several types as well as their intercorrelations in a single image, and the use of false stereo in which any data set (e.g., gravity) can be merged as a stereo component of the image of any other data set. The basic goal is the evoluation of techniques for the presentation of immense multivariate maps or data sets directly to the human brain, by far the most sophisticated data interpreter.
