M
M
Mirex19112016-02-15 02:50:45
CAPTCHA
Mirex1911, 2016-02-15 02:50:45

How can I recognize such a captcha using OpenCV?

Here you need to click inside the open circle so that the door opens.
42f2b1b36bf8464a91e6159cce76722c.png 
3ad702d50b664646bab5663819284fcf.png
I can’t even find any mention of this captcha anywhere, and even more so a ready-made solution for bypassing. I decided to start learning OpenCV and remembered this ill-fated captcha. Maybe someone will tell you in which direction to move, how you can determine the center of an open circle. The Hough transform does not detect a circle.
UPD:
Multi-colored figurines are not the best solution for developers. It turned out something like this:
4e79b396ac80409ba50ad8292b7a2d50.png
how can you now identify pictures with a triangle and a rhombus?

Answer the question

In order to leave comments, you need to log in

2 answer(s)
M
Michael, 2016-02-15
@Mirex1911

segment by color, look for circles for individual colors, see if the contour is open, check the location of "opens".
PS: according to the pictures presented, it can be assumed that the open circuit is the brightest, which can greatly facilitate the task :)

M
Mirex1911, 2016-02-17
@Mirex1911

Done :)
1) Segmented by colors
2) Hough's transformation found circles (on the 3rd column)
3) Calculated the average number of pixels per unit area of ​​the square describing the figure. For an open circle, this value is ~2 times less than for a closed one (on the 3rd column)
4) ???
5) Profit!

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question