|
Key Features
- A simple interface and workflow process for extracting a wide range of geospatial features including land cover, hydrology, transportation, buildings, and small manmade objects
- The ability to take into account spatial context (contextual classification) and ancillary data sources such as Digital Elevation Models (DEM)
- Reinforce the extraction model to improve results by selecting correct and incorrect examples from the initial returned set of features. This iteration, called Hierarchical Learning, is an essential component of the Feature Analyst machine-learning process.
- Object-specific change detection
- Robust image processing tools for fusing or manipulating image data
- Advanced 2D and 3D vector editing tools
- Workflow modeling
- Batch Processing
Feature Analyst 5.0 Enhancements
Building Collection Toolkit
The Building Collection Toolkit is a collection of semi-automated tools that are designed to minimize the amount of user input that is required to extract a complete building from RGB and panchromatic imagery. The tool will ingest a single image and will extract building rooftop outline polygons from the image based on an oriented rectangle the user defines around the building.
The user also has the option of collapsing small buildings to points if the buildings are under a user-defined size. In addition, the tool automatically extracts certain building attributes, such as perimeter, area, and orientation, and allows the user to specify their own custom attributes for each building.
With minimal input, the Building Collection Toolkit allows the user to extract attributed, production-quality buildings from RGB and panchromatic imagery in a much shorter amount of time when compared to standard hand digitizing.
Feature Simplification
The new Feature Simplification tool streamlines and automates what used to be a labor intensive process: incorporating Feature Analyst extraction results into an existing database with a predefined schema and stringent collection specifications. Many attributes such as orientation, length, width, and area can now be computed automatically, and collection specifications – such as minimum polygon size, minimum vertex spacing, and collapsing narrow polygons to lines – can also be automatically enforced.
The intuitive yet powerful interface gives analysts quick access to the most commonly used collection specifications, but also ultimate control over defining what must be done to a feature layer in order to incorporate it into an existing database, or modifying it to adhere to an existing set of map specifications. Several common collection specifications are included by default, and analysts can also define and save their own custom specifications, which can then be applied to multiple feature layers.
The Feature Simplification tool puts the computer to work at what it does best: organizing and modifying vast amounts of data, and frees up analysts to do what they do best: high level cognitive analysis and decision making.
Faster Extraction Times
Large images can now be processed in much less time using the improved multi-threading capabilities of Feature Analyst. Feature Analyst can now leverage multi-core systems and gain a near linear increase in processing speed on complex extractions.
Tighter integration with LIDAR Analyst
Feature Analyst and LIDAR Analyst now offer the ability for users to build Automated Feature Extraction models (AFE files) that contain a combination of both Feature Analyst and LIDAR Analyst processes. For example, a user might generate a ground mask using LIDAR Analyst and use this as an input into Feature Analyst to extract various ground based features including roads, sidewalks, or vehicles. This combined model could then be used to batch process similar datasets.
This capability provides users with the unique ability to leverage both imagery and LIDAR data to achieve optimum results.
Learning Settings Discovery
The new Settings Discovery option uses artificial intelligence to automate the process of analyzing and choosing the most appropriate settings for a specific extraction. Analysts can now choose to have Feature Analyst automatically perform what previous versions of Feature Analyst required them to do manually, both saving analyst time and improving the quality of results.
Improved Batch Classification
Users can now batch models that contain Feature Analyst processes, LIDAR Analyst processes, or a combination of both all from the same interface. This model-based tool provides an intuitive user interface for rapidly setting up batch jobs with an emphasis on reducing the number of mouse clicks required to create a batch job as compared to the previous tools.
The new batch supports launching Feature Modeler from the batch interface to change inputs or outputs or make other model changes before continuing to set up the batch operation. In addition, the entire batch configuration can be saved as a batch script and reloaded to allow users to execute commonly run batch jobs much faster.
Integration with ESRI's Model Builder
Feature Analyst and LIDAR Analyst Automated Feature Extraction models (AFE Files) can now be incorporated as a Geoprocessing Tool in ESRI’s ModelBuilder. AFE Models can contain any combination of the various tools offered in Feature Analyst or LIDAR Analyst. This powerful capability allows users to enhance their existing Geoprocessing toolbox to create an end-to-end workflow.
Preview Extraction Results
The FA preview function is an important enhancement that allows the user to apply learning to the visible extent of an image prior to processing the entire image. This gives users an opportunity to evaluate their learning settings beforehand and “preview” results before running them on a larger image. This can save time for analysts by not having to wait for large images to be processed before deciding whether the settings are optimal or not. This enhancement also allows users to decide over what extent the learning settings should be applied to, so that any spectral or spatial variation of desired features within the image can be captured.
Reduced Number of Output Layers
Feature Analyst 5.0 now offers the ability for users to reduce the number of output layers that can be generated in an extraction workflow by allowing users to re-use layers for certain types of processes. Existing processes can detect if they have already produced results and prompt users to overwrite existing results. This enhancement will reduce the number of layers that can sometimes clutter a project file.
Enhanced Masking
Users now have the option to mask by a specific pixel value instead of just black or white, mask by feature layer, or mask by the current visible extent of their data.
Enhanced Aggregation
The aggregation process has been significantly enhanced to allow users to specify both a minimum and maximum size. In addition users can specify their minimum and maximum size in terms of square meters, square feet, or number of pixels.
|
