I've been continuing work on reconstruction by stepping through the sample code from the MaSKS website. I think I've found the real reason my normalization wasn't working -- there are some extra scaling steps that go into finding the fundamental matrix that I wasn't taking, and normalizing my projection matrices wasn't the correct fix like I thought before. So I wrote my own function to do this in OpenCV; the existing function cvFindFundamentalMat() does not scale the points for better conditioning.
If I give my final presentation today, you can download it here.
skip to main |
skip to sidebar
This is the blog for my undergraduate research project in computer vision. For more info on the class, click here to visit our course website.
No comments:
Post a Comment