My researches during my college life.

Combine pictures by using Sift

Progress:

Stage 1:Finished

Translate and understand Lowe's paper

Stage 2:Finished

Compare sample code to his paper, debug and understand his code

Stage 3: Under Process

Write my own program base on hisSample, see why and how it works.

Stage 4: not started yet

Improve his code and add surf-sift algorithm to the sample code.

Detail:

Papers in Stage 1:

[1] D. G. Lowe. distinctive image features from scale-invariant keypoints. International Journal of computer vision, 2004.

[2] Y. Ke and R. sukthankar, computer vision and pattern recognition, 2004

Code Used in Stage 2:

Since there is no MATLAB in my computer lab. I will useRob Hess's code instead of David Lowe's.

The code I started to learn was programmed by Rob Hess.

Opencv and GSL library must be installed to compile the code.

Original code can be downloaded here:Http://web.engr.oregonstate.edu /~ Hess/

Program in Stage 3:

Main Feature of the program:

Find relevant areas in each picture, combine them to one image.

I have CT the final program can combine several images captured by cell phones or cameras to a 360 degree image like Google Street View does.

Current progress:

Combine several images to one:

Img1:

 

Img2:

 

Img3:

 

Img4:

 

After combined:

 

 

Temporary output: