Problem with ppi (pixel purity index) function

I have got different first three endmembers when I run ppi(data,3) versus ppi(data,5).

At the same time, according to the function description, the endmembers must appear in the descending order of their PPI counts.

So, the first three should be the same.

Could you please clarify this issue?

Thank you.

Answer 1

[Image Analyst]

Here is the help explanation:

------------------------------------------------------------------------------------------------------------------------------------

endmembers = ppi(inputData,numEndmembers) extracts endmember signatures from hyperspectral data cube by using the pixel purity index (PPI) algorithm. numEndmembers is the number of endmember signatures to be extracted using PPI algorithm.

The function projects the hyperspectral data onto a set of randomly generated unit vectors. The pixels with extreme values in the direction of an unit vector are considered pure pixels and they constitute the endmembers. The value of an endmember across all the spectral bands in the input data comprises the endmember signature. For more information, see Algorithms.

Pixel purity index (PPI) method computes the orthogonal projections of hyperspectral data values on a set of randomly generated unit vectors known as the skewers. Then, the method computes the PPI count for each data value. PPI count is the number of times a data value results as an extrema point when projected on to these skewers. Those data values with more than expected number of PPI count comprise the endmembers of the hyperspectral data. PPI is a non-iterative method and the steps involved are summarised as follows:

1. Compute principal component bands and reduce the dimensionality of the input data by using MNF or PCA. The number of principal component bands to be extracted is set equal to the number of endmembers to be extracted.
2. Generate k number of skewers of length same as the input data.
3. Let r be the sample vector that denote a pixel spectra. Then, orthogonally project the sample vector onto each skewers and find the extrema.
4. Store the location of each extreme value and count their occurrences. The number of occurrences is known as the PPI count.
5. Find the PPI count for each pixel spectra in the input data cube.
6. Arrange the pixel spectra in descending order of their PPI counts and identify the first n number of pixel spectra in the ordered set as endmembers. The number of endmembers to be selected is specified by the input argument numEndmembers.

------------------------------------------------------------------------------------------------------------------------------------

So since you're specifying different number of principal component bands to be used (3 and 5) I'm not sure why you expect the endmembers to be the same, especially when the skewers are "a set of randomly generated unit vectors"