Fast High Dynamic Range Image Deghosting for Arbitrary Scene Motion

Fast High Dynamic Range Image Deghosting for Arbitrary Scene Motion

Graphics Interface 2012

 

Simon Silk                                    Jochen Lang

 

School of Electrical Engineering and Computer Science

University of Ottawa

 

Abstract

High Dynamic Range (HDR) images of real world scenes often suffer from ghosting artifacts caused by motion in the scene. Existing solutions to this problem typically either only address specific types of ghosting, or are very computationally expensive.

We address ghosting by performing change detection on exposure-normalized images, then reducing the contribution of moving objects to the final composite on a frame-by-frame basis. Change detection is computationally advantagous and it can be applied to images exhibiting varied ghosting artifacts. We demonstrate our method both for Low Dynamic Range (LDR) and HDR images. Additional constraints based on a priori knowledge of the changing exposures apply to HDR images. We increase the stability of our approach by using recent superpixel segmentation techniques to enhance the change detection. Our solution includes a novel approach for areas that see motion throughout the capture, e.g., foliage blowing in the wind.

We demonstrate the success of our approach on challenging ghosting scenarios, and that our results are comparable to existing state-of- the-art methods, while providing computational savings over these methods.


 

Welcome to the supplementary material page for our paper Fast High Dynamic Range Image Deghosting for Abitrary Scene Motion, which will be presented at Graphics Interface 2012 in Toronto, Canada.

Data sets

The data sets used in our paper are posted below. Each is in a separate zip file. The files are in ppm format (IrfanView is a good free viewer) and includes the exposure data in the form of a hdrgen file of the same format used by pfscalibrate (see here for a description of that file format and how to use it in pfscalibrate).

Marion Lot data set

 

Code

We have released our MATLAB code for this paper. To try it out, grab a copy, and a data set from above. It uses SLIC superpixels, by Achanta et. al., via a MEX wrapper; for this, we currently include only precompiled MEX binaries for 32-bit and 64-bit Windows, but we hope to include source code for that component soon.

Source Code

 

Paper

S. Silk and J. Lang, “Fast High Dynamic Range Image Deghosting for Abitrary Scene Motion,” in Proc. Graphics Interface, 2012.

 

BibTeX:

@INPROCEEDINGS{silk2012fast,
  author = {Silk, Simon and Lang, Jochen},
  title = {Fast high dynamic range image deghosting for abitrary scene motion},
  booktitle = {Proc. Graphics Interface},
  year = {2012},
}