Finding rotation angle inside a license plate

Question

I am trying to correct the distortion inside license plates such as:

However, I can't find a robust way to detect the rotation angle. I was trying to use eigenvalues as here but it fails.

I was also thinking about Hough lines detection but results remain poor.

What can I do to improve the rotation detection?

OK. Can you also upload the segmentation mask for this image? — Nouman Riaz Khan, Dec 20 '18 at 12:43

score 1 · Accepted Answer · answered Dec 29 '18 at 07:45

Find plate contour and then find the angle of the plate by cv2.minAreaRect

#preprocessing steps
...
#find angle
im2, contours, hierarchy = cv2.findContours(preprocessed_sloping_plate,cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE)

#contour with the largest area is possibly the plate
max_area = 0
max_cnt = None
for cnt in contours:
    area = cv2.contourArea(cnt)
    if(area > max_area):
        max_area = area 
        max_cnt = cnt

min_rect = cv2.minAreaRect(max_cnt)
(x,y,w,h,angle) = min_rect

#rotate
M = cv2.getRotationMatrix2D((w/2, h/2), angle, 1.0)
rotated_plate = cv2.warpAffine(preprocessed_sloping_plate, M, (w,h))

Finding rotation angle inside a license plate

1 Answers1