IRMAD service specifications

This service provides a binary change detection map from calibrated products of the same optical or radar sensor, using the Iteratively Reweighted Multivariate Alteration Detector (IRMAD) algorithm.

The tutorial of the IRMAD service is available in this section.

Service Description

The IRMAD Change Detection (IRMAD) processing service provides a binary change detection map from calibrated products of different optical and radar sensors using the Iteratively Reweighted Multivariate Alteration Detector (IRMAD) algorithm (Nielsen, 2007)¹. The Multivariate Alteration Detection (MAD) algorithm (Nielsen and Conradsen 1997)² is a change detection algorithm based on canonical correlation analysis (CCA). The binary change detection map is then obtained via a k-means clustering into 2 classes and using the chi-square distance calculated in MAD³,⁴. In this change detection service the same number of assets having the same CBN shall be taken from pre- and post-event calibrated datasets.

Note

IRMAD requires at least two CBNs from pre- and post-event datasets (minimum 4 assets).

Note

the change detection is applied to all image pixels in the AOI thus clouds and/or water bodies will potentially be part of changes detected. Also change detection results depend on georeferencing accuracy of input pairs of pre- post-event single-band assets. The change detection of the image stack is made after a co-location of input single-band assets without a co-registration.

Inputs

The IRMAD service requires as input a pair of calibrated Datasets from Optical or SAR missions. Ideally, input SAR Datasets to be made with a pair of Calibrated Datasets from the same mission, track, and polarization.

Warning

This minimum number of single-band geophysical assets must be more than one per Calibrated Dataset. Therefore, CVA requires at least two CBNs from pre- and post-event datasets (minimum 4 assets).

Parameters

The IRMAD service requires a specified number of mandatory and optional parameters. Table 1 describes the service parameters.

Parameter	Description	Required	Default value
Input pre-event product reference	Pre-event input product to be used in creating the collocated stack and performing IRMAD.	YES
Input post-event product reference	Post-event input product to be used in creating the collocated stack and performing IRMAD.	YES
List(s) of comma separated bands	This parameter is a list of bands expressed as a comma separated list of common band names. It defines the list of common band names to extract. List of single band assets to be extracted from the calibrated dataset before and after the event. The service requires the same assets in the pre- and post-event calibrated datasets.	YES
Area of Interest	This parameter defines the area of interest expressed as a Well-Known Text value. If set, it overrides the automatic determination of the maximum common area between the input-reference products geometry.	YES
Maximum number of iterations in the canonical correlation	Maximum number of iterations in the canonical correlation (e.g. 30). If set to 1, IRMAD becomes MAD.	YES	30

Table 1 - Service parameters for the IRMAD processor.

Input pre- and post-event product references

The pre-event and post-event Calibrated datasets input products to be used in creating the collocated stack and performing IRMAD.

List-of-comma-separated-bands

This second mandatory parameter is a list of bands expressed as a comma separated list of common band names. The list of single-band geophysical assets to be used for the co-location shall be given as a list of comma separated CBN.

Example

To define multiple reflectance single-band assets from VIS and NIR (e.g. blue, green, red, and nir) from pre- and post-event Calibrated Dataset, the user shall define the 4 input assets in IRMAD as following:

blue,green,red,nir

AOI

This parameter defines the area of interest expressed as a Well-Known Text value.

Tip

In the definition of “Area of interest as Well Known Text” it is possible to apply as AOI the drawn polygon defined with the area filter. To do so, click on the button in the left side of the "Area of interest expressed as Well-known text" box and select the option AOI from the list. The platform will automatically fill the parameter value with the rectangular bounding box taken from the current search area in WKT format.

Maximum number of iterations in the canonical correlation

This last parameter the user shall define the maximum number of iterations in the canonical correlation of IRMAD (e.g. 30).

Note

If the number of iterations in the canonical correlation is set to 1, IRMAD becomes MAD.

Output

The result product of the IRMAD service is a single-band change detection binary map (1 change, 0 no change) GeoTIFF in COG format. Product specifications for this service are shown in the below Table.

Product A: Change Vector Analysis binary mask product

Attribute	Value / description
Long Name	IRMAD change detection map
Short Name	irmad_change_detection
Description	Binary change mask: 0=No-change, 1=Change
Data Type	1-bit
Band	Single
Format	COG
Projection	Native or EPSG:4326 - WGS84
Units	N/A
Valid Range	[0 - 1]

Filter and or Vectorize IRMAD change mask single band asset

IRMAD's binary change mask asset can be spatially filtered and / or converted to polygon using the FilterVectorize service.

To further post-process the irmad_change_detection single band asset by removing small isolated clusters of pixel employ the FilterVectorize service in Filter mode by selecting a filter threshold size value.

The binary change mask can also be converted to polygons by using the FilterVectorize service in Vectorize mode and selecting only true values DN=1 (change).

To apply both spatial filtering and vectorization on IRMAD's binary change mask employ the FilterVectorize service in the Filter and Vectorize mode.

Warning

Only the irmad_change_detection single band asset can be used in the FilterVectorize on-demand service, being the only discrete raster produced by the CVA service.

A. A. Nielsen, “The regularized iteratively reweighted MAD method for change detection in multi- and hyperspectral data,” IEEE Trans. Image Process., vol. 16, no. 2, pp. 463–478, 2007. ↩
Nielsen A. A. and Conradsen K. (1997), “Multivariate alteration detection (MAD) in multispectral, bi-temporal image data: A new approach to change detection studies”, Informatics and Mathematical Modelling, Technical University of Denmark, DTU. ↩
Switzer P, Green A (1984) Min/max autocorrelation factors for multivariate spatial imaging. Technical report no 6, Department of statistics, Stanford University, Stanford, California. ↩
Nielsen A. A., Conradsen K., & Simpson J. J. (1998), “Multivariate Alteration Detection (MAD) and MAF Postprocessing in Multispectral, Bitemporal Image Data: New Approaches to Change Detection Studies”, Remote Sensing of Environment, 64(1), 1-19. DOI: 10.1016/S0034-4257(97)00162-4. ↩