[yotamarker]

how should I go about making  an algorithm that uses a camera to recognize grids like game boards
and floor tiles ?

[keghn]

 The algorithm is called a "edge detector'. There is the Sobel and then Canny.
 And the edges they detect are feed into Surf algorithm or  Sift algorithm to find features,
like boxes and circles and corners.
  Then the more complicated ones, That need do be trained, are the Haar detector and
the most complicated and best of all is the "convolution neural network".

 These can all be programed in python or c/c++.
 

[yotamarker]

doesn't look like them algs can define grids or cog wheels.

[keghn]

https://en.wikipedia.org/wiki/Chessboard_detection

OpenCV free software:
https://en.wikipedia.org/wiki/OpenCV


 Corner detector algorithm:
http://docs.opencv.org/3.1.0/dc/d0d/tutorial_py_features_harris.html#gsc.tab=0

[yotamarker]

found http://accord-framework.net/docs/html/T_Accord_Imaging_HarrisCornersDetector.htm
and
http://accord-framework.net/docs/html/M_Accord_Imaging_HarrisCornersDetector_ProcessImage_1.htm

for now

[keghn]

How Snapchat's filters work:
https://www.youtube.com/watch?v=Pc2aJxnmzh0&feature=youtu.be