0 Comments
The project focus on feature matching and image stitching. We take two images, apply Gaussian smoothing and find Harris corners. Then, it takes a patch of 32, 32 pixels centered at each corner and applies some similarity measurement algorithm and threshold to get best matched features. I have tried correlation coefficient, cluster reward and normalized mutual information. Correlation coefficient is fast but gives some wrong matches. On the other hand, mutual information (MI) is computationally expensive but gives better results.
The project is implemented in MATLAB 2010. |
Archives
August 2019
Categories
|