I have tested Shi-Tomasi corner detector (http://docs.opencv.org/2.4/doc/tutorials/features2d/trackingmotion/good_features_to_track/good_features_to_track.html) which is great to locate areas with text when you have sharp images. But I want to ask you if is it possible to do something similar one text which is blured.
This is source image:
http://oi67.tinypic.com/vg7u50.jpg
And this is what I get
http://oi64.tinypic.com/vgltp4.jpg
the upper image parameters minDistance=5, qualityLevel=0.01, the lower image minDistance=10, qualityLevel=0.05
the rest of params are same for both images: int blockSize = 3; bool useHarrisDetector = false; double k = 0.04;
I would like to get better results to find areas which are not that sharp as in the bottom right corner, but rather so blured as the number "195" in upper right area. More points to find from the blured bottom left, e.g. the "14 ms" and/or "19 ms" text could be detected? Is there some function which can do that? Maybe you you recommend some tutorial which can do it?
↧