Laplacian Pyramid Image Enhancement

My final project for the Computational Photography class at GaTech.

Multi-scale image contrast/detail manipulation

Operating on images at multiple scales is conventionally done using multi-scale decompositions, such as the Laplacian pyramid [Burt and Adelson 1983]. Such techniques decompose an image to multiple detail layers at different scales in a pyramid like fashion.

This project demonstrates how a laplacian pyramid based on the bilateral filter [Tomasi and Manduchi 1998; Durand and Dorsey 2002] can be used for exaggerating details in an image.

I made a simple program in OpenCL that allows for interactively manipulating images at 3 different detail layers and the color base layer.

 

Media

Read the Class Slides for more explanation!

Watch the Demo Video to see the program in action!

 

Excerpt from my class slides

 

 

Example

Stream - original
Stream – original
Stream - adjusted
Stream – adjusted

 

Download:

PyrDetail – Mac app (64 bit)
PyrDetail – Linux app (64 bit)

Note 1:
These libraries required to run the app on linux:
* OpenCV
* Qt  (OpenCV installs some Qt libs, but just in case)
* OpenCL  (should already be installed by your graphics card driver)

Note 2:
The Mac version is more up-to-date (with added pre-smoothing option), and should now run as a drag-n-drop App!

 


Update: 03/15/2013
New version 0.7
-Added presets
-Added histogram
-Misc internal tweaks
-(updated Mac Only, linux update will come soon)

Screen Shot 2013-03-29 at 12.41.46 PM

Screen Shot 2013-03-29 at 12.42.26 PM

 

(SEO laplacian pyramid image enhancement)

1 thought on “Laplacian Pyramid Image Enhancement”

Leave a Comment