WP2 HUEX tools webpage

logo

HUman EXposure to multiple risks (HUEX) Tools


Original HUEX Processing chain

The design of the processing chains useful for the HUEX tool is reported here.

Processing chain: Human Settlement Extraction

chain1

Processing chain: Human Settlement Extraction

SPT chain

HUEX Tool implementation

The implementation of the first version of the HUEX tool focused on developing of a generic method for extraction of human settlements at the global level. The generalisation goal related to the fact that depending on the geographic locations and particular cultures around the world, the concept (or definition) of a human settlement may vary to a large extent.

The method was then devised in a way that it can be adapted to adhere to different human settlement definitions. Moreover, the method can cope with different sources (sensor systems) of remote sensing image data (e.g., optical, SAR, hyperspectral).

The processing chain starts by searching for hypotheses of settlements in remote sensing images, using the Urban Focus operator (an original component of BREC). Then, image segmentation followed by a classification procedure, performed in InterIMAGE, identifies individual buildings within the tentative settlement region. Depending on the particular definition of a settlement, stated in a knowledge model with user defined parameters (e.g., density of buildings), the initial settlement hypotheses are discarded or validated.

In the first version of the human settlement extraction processing chain, BREC functions an InterIMAGE were executed in an offline fashion.

chain1

The second version of the HUEX tool focused on the integration of the BREC algorithms in InterIMAGE, so that a single software suite is needed to implement the whole processing chain is the InterIMAGE system.

In a parallel development, a batch mode of operation was included in the InterIMAGE system in order to ease the processing of multiple data sources with the same knowledge model (for human settlement extraction).

During the TOLOMEO Project yet another useful software tool, the Segmentation Parameters Tuner (SPT v. 3.0), was implemented to provide support for the segmentation task within the settlement extraction procedure implemented by the HUEX tool.

SPT helps the users of the HUEX tool to cope with three important issues: (i) the large variety of remote sensing image sources available; (ii) the variability of human construction patterns in different regions around the globe; and (iii) the different segmentation algorithms available. SPT’s main task is to find optimum segmentation parameter values, for a specific segmentation algorithm from the ones available in the InterIMAGE suite, and for a specific image type. The user provides SPT with segment references; the software in turn searches for proper segmentation parameter values so that the automatic segmentation procedure generates segments as similar as possible to the references. SPT also offers a choice of optimization procedures for the search procedure.

Additional documentation

The final presentation of the HUEX tools and their results, provided during TOLOMEO Final Meeting, is available here. Moreover, a video tutorial of the tool is available here. The video tutorial for SPT can be found here. Finally, the so called “protocol documents”, showing how to use the tool in practice, can be obtained here (for the on-line operation mode), here (for the batch operation mode); and here (for SPT).

Where can one find and download HUEXT Tool?

Prebuilt installation programs and the source code of the InterIMAGE software suite (already featuring the integration with the BREC algorithms) can be downloaded from: here (make sure to download the latest version).

Specific information about the operators presented in the video tutorial can be found here.

If one needs help with the InterIMAGE software, there are a mailing list, a wiki and a user guide here .

The SPT software can be downloaded from here.

Documentation about the SPT Project as well as the software’s user guide can be found here.