There is an image captured by camera, and I need to find the angle of the text in order to rotate it to make the image better for OCR results.
So I know that the fourier transform can be used for that purpose,
My question is, does it really gives good results or may be it is better to use something different than that?
Can you tell me if there is a good method for this purpose?
I am afraid that not every image containing the text will give me a good result after using fourier transform method.
Actually, if I make like it is written here:
link text
(see the part related with an example of text image)
calculating the logarithm of the magnitude of the Fourier transform of image with text
and then thresholding it, I get that points and I can calculate the line approximately passing through them, and after getting the line calculate the angle, and then make an affine transform,
But, what if I do not get a good result every time using this method , and make a false transform?
Any ideas please to judge wether the result is correct or not, or may be another method is better?
The binary image can contain noise, even if there are not so much of them, the angle found as a result can be not accurate.