Regionprops on binary image gives wrong values

I am trying to using region props to detect some blobs in a binary image. There is very little noise in the image but when I run region props and look at the 'stats' output I get a very long array of many small centroids also the first row of the stats output is the only large value for in the array, about 2000 in size when the image is only 2048x2048. Here is the output: https://postimg.org/image/p2shlqn8b/

Is there something I am missing about region props? I thought it should easily return the centroids and axes. Here is the image I am using and I run this line of code on the image BW https://postimg.org/image/sse8sobzp/ stats = regionprops(BW,'Centroid','MajorAxisLength','MinorAxisLength');

Thanks for your input!

[Image Analyst]

I'm sure there's something you're missing or not understanding. Please post your images here with the green and brown frame icon (preferred), or the paper clip icon, along with your m-file, rather than on some third party, external site that we'd have to go to. You can do it all here.

[Joe Parks]

Thanks for the advice, here is the image and the mat file containing the arrays.

[Joe Parks]

Nevermind, I figured it out. The image needed to be inverted with

BW = ~BW;

Answer 1

[Image Analyst]

Yes, the "foreground" is the white blobs and the "background" is the black blobs. bwlabel() labels only the foreground regions and regionprops() measures only the foreground regions.