Best approach for a specific Object / Image Recognition task?

I am looking for a specific object in my photo:

Object: the outline of a rectangle with an X in the middle. It looks like a rectangular flag. All this. So, do not fill, just lines. The rectangle will have the same length ratio in width, but can be of any size or any rotation in the photo.

I looked at a bunch of image recognition approaches. But I am trying to determine the best for this particular task. Most importantly, the object is made of lines and is not a completed form. In addition, there is no distortion of perspective, so a rectangular object will always have right angles in the photo.

Any ideas? I hope that I can implement quite easily.

Thanks to everyone.

+5
source share
3 answers

You can try using an angular detector (such as Harris) to find the corners of the window, the ends and the intersection X. This makes it easy to find points in the correct configuration.

Edit (response to comment):

I assume that you can find the corner points in your image, 4 corners of the rectangle, 4 lines of the end of X and the center of X, as well as several other corners of the image due to noise or objects in the background. This simplifies the task of finding a set of 9 points in the correct configuration from a given set of points.

, A. B A. , , () A - , B - , , , . ( , FLANN), , , . , , , , .

, . ( , + X), , ( ), . , , . , , / , .

+3

hough. , , . . - , .

+1

. -

, - , , , , , . ( , ) .

: :

0

All Articles