0

I'm writing c++ image processing project using opencv. i'm calculating variance of image for each pixel. You may say variance filter. For each pixel i calculate variance around that pixel with filter size. FilterSize much smaller then image size. For that i used blurring:

    blur(workingIm, blurredImage, Size(filterSize, filterSize));
    blur(workingIm.mul(workingIm), blurredSqureImage, Size(filterSize, filterSize));
    sqrt(blurredSqureImage - (blurredImage.mul(blurredImage)), varianceImage);

i try to improve performance using integral image. When i write loop of my own it was slower then original.

Are there any build-in opencv functions to speed up the process? Or how can i write loops that can catch opencv performance?

fyo
  • 77
  • 2
  • 13
  • It's actually not very clear what you're doing, or what you're trying to do. Any reason why you don't use [meanStdDev](http://stackoverflow.com/a/32967631/5008845), squaring the standard deviation? – Miki Feb 23 '16 at 10:32
  • I edited question. I look into meanStdDev. As far as i understand it is not suits for my needs. – fyo Feb 23 '16 at 14:36

0 Answers0