Project 1: High Dynamic Range Imaging
Computer Vision – CS766 – Fall 2008

Christopher Hopman

 

Outline

Introduction
Approach
Results
References

 

Introduction

Several algorithms were implemented for this project:

1. Correcting radial distortion
2. Warping an image to a cylinder
3. RANSAC to find a homography
4. Image warping and stitching to form mosaic

 

Approach

Image Capture

Images were captured with a Canon Powershot SX100 IS mounted on a tripod with a Kaidan head. Camera was very poorly aligned, leading to some fun difficulties later on. Camera settings were controlled remotely via Canon's RemoteCapture DC software, and pictures were taken via this software.

 

Radial distortion and cylindrical warping

Basically this is just implement a simple reverse warp... I did two warps, could have been done with one

 

RANSAC

Again, basically just implementing the algorithm. The previous step, finding feature points and
matching them was done with the given SIFT software and code (which was modified a bit).

Warping and stitching

So, due to my very bad setup of the camera this became rather complicated.
here's an early result.
I believe one thing that also caused me problems is the huge difference in distance from the camera
as you go from one part of the mosaic to the other.

Results

The mosaic results for my images were not that good. As noted above I think that several things led to less than stellar mosaics. Here's some pics

I have not been able to get my own images to turn out well. Below are some results on the sample images.

Anyway, some of the code stuff. Basically every step is coded seperately. They take an input file that is generated by the step before. Do what they do, then you run the next one.

oh, and a result on something a little bit better behaved.

and that one with the error corrected. error seemed to accumulated with square of distance so
corrected by that, rather than just linearly. Links to the viewer with this image.