License plate detection question

i was trying to implement an ANPR algorithm in matlab. I will try to follow this simple algorithm proposed by rodrigoap in comments

What are good algorithms for vehicle license plate detection?

first I applied the Sobel operator and then masked all the image with an image of a plate (with the Sobel operator applied too). The region of maximum coincidence is where the plate is. Then apply an OCR to the selected region to get the number.

my question is: what does masking an image mean?

does it mean to convolve it with the whole image or does it mean to simply look for the portion of the image where the difference betwe开发者_如何转开发en the subimage and the test number plate is minimum?

your help will be highly appreciated. Thanks

It's difficult to know what he meant by "masking", but one method that could work is using normalized cross correlation, which is a measure of the similarity between an image and a subimage.

If you have the image processing toolbox in matlab, I suggest you use the normxcorr2 function. This function takes a subimage and a image as inputs and gives out a correlation image. The peaks of the correlation image are located where the two input images are most similar.

http://www.mathworks.com/help/toolbox/images/ref/normxcorr2.html