Subtracting background marks shadows as foreground

I have background subtraction to work quite well. But the problem is that it also denotes shadows as in the foreground.

Has anyone found some technique to get rid of the shadows from the background? Links to technology, documents, articles, etc. Which you used and received reliable, will be amazing!

+6
image-processing opencv computer-vision background-subtraction shadow-removal
source share
3 answers

This is a well-known problem, and, like many well-known problems, it has a lot of research, which means that there is no real "optimal" solution.

Google Scholar offers many documents to solve the problem. The work of Elgammal, Davis and others is a good place to start.

You may also be interested in later work on Ground Shadow Detection in outdoor consumer photography .

+9
source share

Good docs:

N. Martel-Brisson, A. Zakkarin, “A Moving Shadow from the Shadows of a Gaussian Mixture,” International Conference on Computer Vision and Pattern Recognition, CVPR 2005, San Diego, California, USA, June 2005.

N. Martel-Brisson, A. Zakkarin, “Moving shadow shadow detection from a Gaussian mixture shadow model: a step towards robust detection and segmentation algorithms in computer vision,” Canadian Intelligent Systems Annual Conference, IS 2006, Victoria, BC, May 2006 g.

N. Martel-Brisson, A. Zakkarin, “Study and removal of shadow with a shadow through a multidisciplinary approach”, “Transactions on the analysis of samples and machine intelligence”, Volume 29, No. 7, pages 1-13, July 2007

S. Huang, R. Wu, "Uncontrolled Shadow Detection Algorithm," International Computer Symposium, ICS 2008, 2008.

S. Huang, R. Wu, “An Internet-Based Learning Method for Shadow Detection,” Pacific-Rim Symposium on Image and Video Technology, PSIVT 2010, Singapore, Singapore, November 2010

Yours faithfully! Thierry

+4
source share

This is a great article. You can find everything about various modern background subtraction algorithms:

A complete overview of background subtraction algorithms evaluated using synthetic and real videos

Section 4.3.4.3. In this article, you can see a comparison of the algorithms when they encounter varying degrees of shadow in the test video. As a quick conclusion, the MultiLayerBGS algorithm is superior to other methods in test cases with a significant amount of shadow.

You can find a good implementation of this method and more than 35 Andrew Sobral background subtraction methods on this page .

0
source share

All Articles