segmentation license plate
I have this image as shown below. The 'k' is somehow not detected because it is connected to the background. Is there any way to detect the 'k'
I have this image as shown below. The 'k' is somehow not detected because it is connected to the background. Is there any way to detect the 'k'
Asked: 2015-08-03 15:56:12 -0600
Seen: 1,271 times
Last updated: Aug 03 '15
Tricky image segmentation in Python
RGB2Gray::operator() causing segmentation fault when using Python
Face-detection/recognition + segmenting + tracking Difficulty
word segmentation using OpenCV
pyramid_segmentation.py crashes with any different input image
How to segment a part of any object for counting purpose as per given binary image?
Grabcut and foreground image extraction
May be you can use rule for license plate. You have found six symbols and one is missing. So add a rectangle on left using mean distance between found rectangle. Your symbol is here. (sorry for my bad english)
this is only one of the test samples. if i implement this, all others will get affected as well.
It depends if you have got many country license plate or not. Another way is to use all rectangles to fit based line of your letter. After fill outside with your background color.
If your LPs have all the same number of symbols, then you can do some kind of research the characters (or search for the missing ones) based on the found ones, by cutting the the top and bottom parts. This cut should be done base on the rectangles you have (like choosing the mean of the y coord, or the min, you should test to see what is the best); then do the same for the bottom line (based on y and height)
Or, if you have the LP, then you can detect the text line, by doing projection on y axis. there should be 2 minimums: one for the top white part and one for the low white. You can improve it if you do the projection on the binary image (after thresholding)