0

I'm currently doing a computer science masters project creating a parser/interpreter for the turtle/LEGO language. Where you enter in commands forward to draw a line forward(FD) and left turn(LT) and right turn(RT) to change the direction you draw this. Anyway i was looking for an interesting extension and had the idea of using open cv to recognise shapes (like a square or triangle) drawn in black and white. Generating x,y coordinates for it and ideally outputting the commands needed to draw that. For example a square would be:

FD 10 LT 90 FD 10 LT 90 FD 10 LT 90 FD 10

I have experience with C but none really with open CV. How difficult do you think this would be and with a week before hand in do you think it's realistic. Thanks! Albie

aheigins
  • 2,574
  • 3
  • 19
  • 26
  • Image recognition in a week is not very likely, unless you already have a lot of experience using open CV. [This post](http://stackoverflow.com/questions/10196198/how-to-remove-convexity-defects-in-a-sudoku-square) may be of interest to you. – user3386109 Jan 21 '15 at 00:33
  • Do you have any experience with the C++ language? It's not the same as C, for example, C++ has classes and the `std::string` type. Which language will you implement it in (choose your tags wisely). – Thomas Matthews Jan 21 '15 at 01:28
  • Thanks for the link, that's really impressive, although the project needs to be written in C, sadly we cant do c++ for this module. I also think it might be able to be simplified down, to only allowing images which are drawn in black and white, and with clear intersections, if i can recognise the location of the points and the length of lines connecting them i can look at working back from this to code, do you think this is doable? – aheigins Jan 21 '15 at 01:35

0 Answers0