Computer Vision Notes
- OpenCV
  - Python
  - C/C++
  - Tutorials
- Ideas
  - Marbled Paper
  - Other stuff
- Great Tutorials
- Libraries
- Classes
- Embedded
- Simple
- Books
- Face Recognition/Projection?!
  - Tracking

Computer Vision Notes

How to solve sudoku puzzles (convexity and OCR)
How to detect circle dots: Dot Detection on StackOverflow. It's all about the parameters!

OpenCV

Python

Probably use Python? And just use cv2. Here's why

	Yes, you're correct. cv is close to the 1.x OpenCV API, and cv2 to the 2.x API. And better to use cv2, because cv is going to be legacy. There is a set of samples for both cv2 (opencv_trunk/samples/python2) and cv (opencv_trunk/samples/python). 

	Please also note that OpenCV python API heavily relies on numpy functionality. And many things are supposed to be taken from numpy, you can see it in samples everywhere. Actually there is no OpenCV's Mat in python, numpy's arrays are used for images.
Regards, 
Kirill

For Python interface, you need to install the cv2.pyd file into Lib/site-packages for things to work correctly.

C/C++

http://stackoverflow.com/questions/10860352/getting-started-with-opencv-2-4-and-mingw-on-windows-7
Will need to include a lot of libraries, so just be lazy and include them all?

 C:\opencv\opencv\samples\cpp> g++ -I"C:\opencv\opencv\build\include" -L"C:\opencv\opencv\build\x86\mingw\lib" video_homography.cpp -lopencv_core242 -lopencv_highgui242 -lopencv_imgproc242 -lopencv_video242 -lopencv_features2d242 -lopencvcalib3d242 -o video_homography

Tutorials

Sudoku OpenCV http://sudokugrab.blogspot.com/2009/07/how-does-it-all-work.html
ARToolkit Tracker: http://www.hitl.washington.edu/artoolkit/documentation/vision.htm
ArUCo tracker: http://www.sciencedirect.com/science/article/pii/S0031320314000235

Ideas

Marbled Paper

The below marbled cover notebooks have a repeating pattern. Can you find them?

Google search. I have more in Dropbox, and this lady did a study of them!

You have some more in Dropbox/eng/python/examples/cv

Other stuff

Characterize text vs. images based on fractal “roughness”. Variance might work too…definitely shouldn't be localized variance though

Great Tutorials

How to train a Haar Classifier

Libraries

OpenCV *
- The SERVO Magazine tutorials that started it all (Haar Classifiers for faces…very understandable){:programming:opencvpart01.pdf|Part 1}} Part 2 Part 3 Part 4 Part 5
JavaCV incorporates OpenCV plus some other cool stuff (AR, Kinect, etc)

Classes

Johns Hopkins, has some sample projects too.
Brown, plus his computational photography class is good too.
OHSU Biomedical Imaging

Embedded

ChipSight . Uses FPGA to compress pixel data into understandable chunks to save processing power. Neat idea, should come out with product in early 2012.