Physical Significance of Kurtosis in image

2 views (last 30 days)
Hi
I am working on image texture analysis, I am trying to understand image Kurtosis. THere are different ways to interpret the value, Can you guys help me to understand physical significance, what does the value of the kurtosis value tell about a image.
Thanks

Accepted Answer

Image Analyst
Image Analyst on 9 Jan 2012
Kurtosis is how "box-like" the histogram is. A kurtosis greater (I think) than 0 has a rather flat top and steep tails that are far away from the middle, king of like a rounded rectangle. A kurtosis less than 0 has a histogram that is rather spike-like with a tall central spike and long, low, rather flat tails.
A positive kurtosis image would have a fairly uniform distribution of gray levels but not too many at the extreme values, while a negative kurtosis image would have mid-level gray levels that are fairly dominant (that's what the central spike represents) and an otherwise uniform distribution (for gray levels away from the mid-gray level).
See my code for calculating skew and kurtosis in http://www.mathworks.com/matlabcentral/answers/15307-image-operations-skewness-and-kurtosis, along with an adidtional explanation.
  4 Comments
Image Analyst
Image Analyst on 9 Jan 2012
Like I said, and the web page says, the higher the kurtosis the more pixels will have the dominant gray level and fewer pixels with other gray levels, thus it's more uniform and has lower noise. An infinite kurtosis would be a spike (delta function), and you can see it would be zero because the standard deviation for such a distribution is zero, so in that extreme you would have only one gray level and no noise at all.
ram m
ram m on 10 Jan 2012
Thank you for the explanation Sir

Sign in to comment.

More Answers (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!