TNTmips provides a full suite of flexible tools for performing spatial/spectral classifications of your images.
The Automatic Classification process provides automated unsupervised and supervised multispectral image classification with extensive post-classification analysis tools, error checking, and class merging.
The Feature Mapping process lets you visually identify class sample areas to guide an incremental image classification.
You can analyze and classify hyperspectral images in the Hyperspectral Analysis Process.
The Raster Unique Value Combination process lets you combine classification results from different dates or conditions for analysis.
See also: Image Classification documentation with tutorial exercises.
Classification restricted
using mask
Image Classification Highlights
- Work with large images (full scenes, scene mosaics) and multiple images from different sensors and/or dates
- Supports mixed input (composite-color, single-band, different data types)
- Allows scaling adjustments to input cell value ranges
- Restrict classification to desired areas of any shape using mask
- Automated unsupervised and supervised classification
- K-Means, ISODATA, Maximum Likelihood, and many other classifiers including neural network methods
- Automatic display of classification result allows comparison with source imagery or any other geospatial data
- Graphical presentation of class statistics (dendrogram, scatterplots, cooccurrence matrix) to aid interpretation and analysis
- Reload classification result at any time for more analysis and modification
- Progressively merge classes with unlimited undos
- Name classes and save interpreted classification results at any time
- Create training sets for supervised classification manually or from attributes of polygon or point data
- Separate statistics compiled for training set and for classification result
- Error matrix shows accuracy of supervised classification
- Use your visual interpretive skills to guide an incremental classification of your image (Feature Mapping process)
- Classify hyperspectral images and perform subpixel spectral identification
- Combine classification results for different dates or conditions into a single image showing all combinations of classes to determine correlations between different spatial conditions or detect change through time
