How to filter out outliers images from a set of images

asked 2017-09-10 01:01:14 -0500

I have a set of images to be used for background subtraction. These images are continuously taken using a fixed camera. The scene sometimes got exposed to a very powerful sunlight, or sometimes affected by some dark shadows.

I am looking for a method to exclude these outlier images from being used for model building.

I am using c++, opencv 3.3.

edit retag flag offensive close merge delete

Comments

try to use meanstddev functions to filter images : low mean and low stddev or high mean with low stddev are not good candidate for background subtraction.

Now low or high you must find a threshold value

LBerger gravatar imageLBerger ( 2017-09-10 02:39:02 -0500 )edit

Hello @LBerger, These functions are to be used and compare among the same matrix elements, not among different matrices. I guess I am looking for something that involves several images. Thanks for your reply

Abdallah Nasir gravatar imageAbdallah Nasir ( 2017-09-10 02:50:14 -0500 )edit

" I guess I am looking for something that involves several images" may be you should think your problem in another way :

If image mean mean is too low don't add to background subtraction method

or

If image mean mean is too low process image and add to background subtraction method

LBerger gravatar imageLBerger ( 2017-09-10 07:02:10 -0500 )edit