 |
| |
|
| ||
Image Processing in IDRISI TaigaRemotely sensed imagery is an excellent resource for land cover mapping and the detection of land cover change or for suitability mapping and environmental management. IDRISI includes the largest suite of supervised and unsupervised classification techniques in the industry, based on scientifically proven algorithms and methods, for both multispectral and hyperspectral imagery. The IDRISI software includes all of the general purpose and advanced processing tools required to prepare your satellite imagery at an extremely affordable cost. Although IDRISI provides an extensive suite of image processing tools, what makes the software critical for today’s analysts is that image processing data can be completely integrated with IDRISI's equally extensive set of raster GIS tools, saving effort, costs and resources. Distinctive Image Classification Features
IDRISI Taiga GIS & Image Processing Brochure IDRISI Taiga GIS & Image Processing Technical Specifications
The SEGMENTATION module creates an image of segments that have spectral similarity across many input bands. The image on the left uses a larger similarity threshold than the one on the right, resulting in more generalized, less homogeneous segments. Using this threshold, the image allows for segments that wholly contain building objects.
Classification Tree Analysis is a type of machine learning classifier. Procedures are included for training and pruning a classification tree. This module produces both hard and soft classified maps. There is one soft map for each class associated with the degree of membership for that class at a particular leaf in the tree structure.
A variety of machine learning classifiers are available within IDRISI. Neural network classifiers include a multi-layer perceptron, self-organizing map, and fuzzy ARTMAP. Each allows complete control over all parameters.
Linear spectral unmixing, or linear mixture modeling, is available in IDRISI and provides a means for sub-pixel evaluation. Three options are provided: the standard unmixing method, a probability guided method, and an exhaustive search method.
IDRISI provides a wide range of classifiers for both supervised and unsupervised classification of remotely sensed imagery. Along with ISODATA, as shown here, other unsupervised classifiers include KMEANS and unsupervised routines in both the SOM and the Fuzzy ARTMAP neural network procedures.
The KNN and KMEANS classifiers are just two of the many routines available in IDRISI for image classification. The KMEANS classifier offers three centroid initialization rules: random seed, diagonal axis, and random partition. The KNN procedure offers both hard and soft classification outputs.
|
 |