Image OCR - filtering unwanted data

Question

I mainly read car license plates using Tessract OCR, however, although you can easily select text by changing the contrast, reducing noise, and so on, some parts of the car remain in the image, which causes OCR to give bad results.

For example, take

I can change this easily enough, for example:

I want to eliminate the edges from each plate, here is another example:

I could remove the edges using the pixel manipulation algorithm, however, I do not think it is the correct method and will cause quite a few problems.

I use the following application to test various methods, such as morphology and elimination of unwanted data, so far I have not been successful.

http://www.codeproject.com/KB/GDI-plus/Image_Processing_Lab.aspx

-, , , , .

Answer 1

.

.....

Answer 2

. , . , , . "/", .

, .

, , , . , , .

, :

IsPixelAboveThreshold ?

, , , , .