@sven - Is bulging a filter in CB ? can it help solve recaptchas solving rate % ?
Hi Sven, is Bulging a filtertype in CB?
have you ever thought of adding a bulging filter on CB to test results ?
it's clearly how recaptchas are bulged by dropping a big 3D sphere either to spread the image outwards or inwards (3D sphere, mountain or ditch)
have you ever thought of adding a bulging filter on CB to test results ?
it's clearly how recaptchas are bulged by dropping a big 3D sphere either to spread the image outwards or inwards (3D sphere, mountain or ditch)
Usually by reverse bulging the image where there's a curve it will make the image more at a normal line.
Allmost all recaptcha images have curves
Allmost all recaptcha images have curves
If you reverse bulge where the bending is in text you can stretch it out to be a more straight text.
I couldnt do the middle part cause there were 4 parts loaded in this image and i would either bring the all towards each other or further appart, but like you see in the corners.. I think it could give other results % if this filter is used properly.. but I dont know how far the CB system works and if it is capable in finding those dents in recaptchas to flatten them out like this ? or do you think this is too much for a system and only capable by human interaction.
anyways just shooting an idea to help solve those nasty things...
I couldnt do the middle part cause there were 4 parts loaded in this image and i would either bring the all towards each other or further appart, but like you see in the corners.. I think it could give other results % if this filter is used properly.. but I dont know how far the CB system works and if it is capable in finding those dents in recaptchas to flatten them out like this ? or do you think this is too much for a system and only capable by human interaction.
anyways just shooting an idea to help solve those nasty things...
Comments
For that I have coded two special filters that will de-cursive it and also normalize the height/width of it. That works good but the problem is (as with so many other captchas) that chars are all merged. That makes it hard for any OCR or other algo to determinate what part belongs to what char.