OpenCV has an extensive set of image processing and vision algorithms. Everyone will have their own pet list of things which they would like to see in this. Some of mine "requests" are
- Level Sets; even a base level implementation; there are a number of libraries out there, integration will be a huge headache/challenge!
- Colour Edge detection, without taking recourse to grayscale conversion. Maybe DiZenzo's algorithm could be a used.
Regarding the general development of OpenCV and computer vision algorithms in general, I think that research groups should be more proactive in contributing code, rather than issue MATLAB dll which would work only on certain versions. After a ll, knowledge is for sharing and it would also increase the citation count if the code were available. Rather than churning out paper after paper using a mysterious black code which is available only to the select coterie, it would be nice if the code were freed up. After all, publishing in academia is about sharing knowledge. If it were an industry code, then perhaps the information hiding is perfectly understood.
No comments:
Post a Comment