ERDAS IMAGINE

[Debbie]

Can somebody advise me of how I can perform change detection and then generate a matrix table (post-classification change analysis i.e. 'from to' change) with Landsat classified (thematic) images in ERDAS IMAGINE 2016.

[jdooley · Accepted solution]

Hi Debbie,

This is a two part inquiry, one is an inquiry for performing change detection and the other is an inquiry for using a matrix table for change analysis.

There are varying change detection methods available to create a Landsat classified thematic image in IMAGINE. Two of the more common methods would be supervised and unsupervised classification. There is an abundance of reference material that discusses these methods so I won’t use this forum to present them. IMAGINE offers online references from the Help tab > Reference Library group > Workflows menu > Classification Workflows.

The Hexagon Geospatial Real-Time documentation portal also provides procedural references for classification.

https://hexagongeospatial.fluidtopics.net/

How are you looking to use a matrix table to measure the change analysis? The matrix table could be statistical or spatial. Raster tab > Thematic > Matrix Union creates a spatial matrix raster output. The spatial matrix example discussed below assumes:

• identical landcover categories have been generated for both ‘before’ and ‘after‘ classified thematic images
• the ‘before’ and ‘after‘ raster images are spatially coregistered (same geographical extent) with each other
• the ‘before’ and ‘after‘ raster images have the same spatial resolution (e.g. 30m.)

Before and After landcover  categories.

1. Urban, built-up land
2. Agricultural
3. Field and Rangeland
4. Forest
5. Water
6. Wetland

The 6 x 6 matrix is constructed by where the columns represent the Before image landcover classes and the rows represent the After image landcover clases. The output pixel values represent the spatial coincidence between the pixels of the Before and After classes. There are 36 different possible change classes. For example, matrix output pixel value 4 represents the coincidence of Before class 4 and After class 1. This is interpreted as a change from Forest (4) to Urban, built-up land (1). Classes 1,8,15,22,29,36 represent classes of NO CHANGE between the Before and After images.

Just one example

Regards

Jeff

[Debbie]

Thank you Jeff for the explanation. I have performed supervised classification in ArcGIS and now have an image with 9 landcover classes which I exported as an .img file. However when I try to load the file into Matrix Union I receive the error message "Input files must be thematic"  - can you assist me with this problem

Debbie

[sbent]

Hi Debbie,

Try changing the layer type for your images from continuous to thematic. You can accomplish this by passing your images through the Subset tool. You will not subset the data, but simply change the output layer type to thematic.

1. Open the Subset tool. ERDAS IMAGINE > Raster tab > Geometry group > Subset & Chip menu > Create Subset Image
2. Select one of your classified images as the Input File.
3. Enter a name for the Output File.
4. Change the output layer type from Continuous to Thematic.
5. Click OK to process the data.

Kind regards,

Stephen Bent
Application Support Specialist