As the saying goes: If a worker wants to do a good job, he must first sharpen his tools.
Knowing what to do and how to do it can maximize the use of tools, and in turn study tools.
Recently learned some image processing methods of OpenCV, like blur, binarization, edge detection, etc. What is the effect of these methods in practical applications? There is an intuitive experience through the experiment of license plate recognition.
Welcome to communicate!
#There are a lot of codes on the Internet, so I won't post the code
#The source picture is found on the Internet, if the car owner minds, please contact.
1. Source image
2. Grayscale
Many methods can only handle one-dimensional images
3. Obfuscation
denoise
4. Edge Detection
5. Binarization
form black and white images
5. Close operation
Eliminate small black areas
6. Contour detection
And select the outline of rectangle (license plate) through feature judgment
The picture is to describe the detected contour (red thin line) and the selected rectangular contour (green box)
7. Cyclic identification of the selected rectangular contour
For the recognition of the rectangular content, the above outline recognition process can be repeated, or the cutting can be performed directly.
Experiments are done by contour recognition
-- End --