Licence plate detection.(Removing the shear angle of plate)

Question

Mayur 2011-4-20

1
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/5851-licence-plate-detection-removing-the-shear-angle-of-plate

Hey guys I’m doing a project on license plate recognition (lpr) . I want to remove the shearing of the number plate. I came across a pdf file on lpr . From that file I want to write matlab code for the following part.

A is transformation matrix shown below.
   A= [1   -tan(angle) 0;
       0      1        0;
       0      0         1;];
Let P be a vector representing a certain point in image, such as P =[x, y,1] where x and y are coordinates of that point. The new coordinates  Ps = [xs ys, ,1] of that point after the shearing can
be computed as:
                           Ps = P*A
Let the deskewed image be defined by a function  fs . The function  f s can be
computed in the following way:
  fs (x , y) =  f ( [x,y,1]*A*[1 0 0]T ,[x,y,1]*A*[0 1 0]T )

Here fs represents new image , fs(x,y) represents new co-ordinates of a point in the image and f represents original image on which transformation is carried out. Guys please help me out in converting this part into a matlab code.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Sean de Wolski 2011-4-20

You're going to get a vote for writing a license plate detection question that was not a doit4me; the first of its kind I do believe. Congratulations!

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Jarrod Rivituso 2011-4-20

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/5851-licence-plate-detection-removing-the-shear-angle-of-plate#answer_8169

Hi Mayur,

Have you had a look at imtransform?

http://www.mathworks.com/help/toolbox/images/f12-26140.html#f12-31782

It seems like it has a sheer transform that you could use, which would be significantly easier than trying to write your own code to do the same.

One of the challenges in doing that, as far as I can tell, is that images in MATLAB typically consist of matrices of evenly spaced pixel data, and your transformation might leave you with unevenly spaced data in x and y, so you'd probably have to interpolate or do something like that to make it work.