IMAGINE Discussions

[jenny_h]

Hi guys,

I have to do a subpixel classification and I've got a reference data set which shows me the estimated fraction. I created an AOI with this data and started the procedure. Always the results were not satisfactory. I got only a few single pixels and no whole areas (There have to be more than the classification results show me). Sometimes I got not even pixels in my training set area. I don't know what I've done wrong.

I hope someone have an advice for me.

Kind regards,
Jenny

[sbent]

Output from the Subpixel Classifier MOI Classification function is a single-layer image file that contains the locations of the MOI. The images below illustrate a typical output from the subpixel classifier (yellow, orange and red pixels) overlaying a SPOT image and the ERDAS IMAGINE raster attribute table showing the total number of pixels detected and Material Pixel Fraction for each pixel classified.

There are many factors which could be contributing to your results. I would advise you to review the Subpixel Classifier Strategy and Interpretation and the Subpixel Classifier Workflow sections in the IMAGINE help documentation along with AAI’s Subpixel Classifier FAQs.

Regards,

Stephen Bent

Hexagon Geospatial Support