Depth-Based Selective Blurring in Stereo Images Using Accelerated Framework

Altering the depth of field of an image based on a supplied depth map and user's depths of interest.

Please cite these papers:
[1] S. Mukherjee and R. Guddeti, "Depth-Based Selective Blurring in Stereo Images Using Accelerated Framework", Springer-Verlag Journal "3D Research", vol. 5, no. 3, 2014.
[2] S. Mukherjee and R. Guddeti, "A hybrid algorithm for disparity calculation from sparse disparity estimates based on stereo vision," IEEE 10th International Conference on Signal Processing and Communications (SPCOM), Jul. 2014.
My algorithm generates a blur map from the input depth map and the user's depth(s) of interest. The blur map is then used to perform depth-based selective blurring of the input image using a Gaussian kernel, resulting in an output image with image regions lying in depth(s) of user's non-interest blurred (degree of blurring is controlled by a input parameter), thus effectively altering the depth of field of the input image.
This submission, used in conjunction with my stereo depth map generator available at:
http://www.mathworks.com/matlabcentral/fileexchange/46172-fast-stereo-matching-and-disparity-estimation-by-s-mukherjee-and-prof-g-r-m-reddy

may be helpful especially for amateur / casual photographers using entry-level cameras with limited capabilities for altering the depth of field during shooting. A typical use case is portrait photography.

A similar project done by GOOGLE, although their depth estimation methodology is different:

http://googleresearch.blogspot.in/2014/04/lens-blur-in-new-google-camera-app.html