Maximum Likelihood classification in ArcGIS

[KarolinaKorzeniowska]

I found that in ArcGIS 10.3 are two possibilities to compute Maximum Likelihood classification:

1. Spatial Analyst > Multivariate > Maximum Likelihood Classification​

2. Spatial Analyst > Segmentation and Classification > Train Maximum Likelihood Classifier (and later) > Classify raster​

Is there some difference between these tools? I compared the results from both tools and I have not seen any differences.

[DanPatterson_Retired]

the new links are

Maximum Likelihood Classification—Help | ArcGIS for Desktop  and

Train Maximum Likelihood Classifier—Help | ArcGIS for Desktop and this is of use

How Maximum Likelihood Classification works—Help | ArcGIS for Desktop

or for a search on all topics

Search Result | ArcGIS for Desktop

Now the question is how did you compare? visually? that question is not clear

[KarolinaKorzeniowska]

I compared the resultant maps using raster calculator. I subtracted results of "Maximum Likelihood Classification" from "Classify Raster", the subtraction map had only zero values.

[DanPatterson_Retired · Accepted solution]

They produced the same results because the second link describes the intervening step to get to the classify raster state.

• To complete the maximum likelihood classification process, use the same input raster and the output .ecd file from this tool in the Classify Raster tool.

Were you expecting a different outcome?

[KarolinaKorzeniowska]

Thank you for explanation. I am not expecting different outcome. I am only asking if these two tools have different outcome. If these two tools are doing the same process, for me it is not logic to provide the same tool under two different names.

[AdamBorczyk]

Late to the party, but this might be useful while scripting - eg. you train the classifier one one 'master' image and then apply it to every other image instead of having to compute classes for main image as well. Not a serious difference, but this might be it.

Pozdrawiam

[BRENNEJM]

I realize this thread is old and marked as solved, but it should be noted that the original question does point out two completely separate workflows with different tool inputs.

Workflow one is Create Signatures and then run Maximum Likelihood Classification.

Workflow two is Train Maximum Likelihood Classification and then Classify Raster.

Workflow two is the basic classification which gives the user little control over the output, while workflow one allows the user to set a reject fraction and account for a priori knowledge. Workflow one also allows the user to output a confidence raster.

As is noted in the approved solution, the classification method is identical, it is only the output which is different.