import cv2

import numpy as np

def find_and_crop_plate(image):

# Convert image to grayscale

gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

# Apply bilateral filtering to reduce noise while preserving edges

gray = cv2.bilateralFilter(gray, 11, 17, 17)

cv2.imshow("Original Image", gray)

cv2.waitKey(0)

# Apply adaptive thresholding to isolate plate characters

thresh = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY_INV, 11, 2)

# Find contours in the thresholded image

contours, _ = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

# Find the largest contour (assuming it's the license plate)

largest_contour = None

largest_area = 0

for cnt in contours:

area = cv2.contourArea(cnt)

if area > largest_area:

largest_contour = cnt

largest_area = area

# Check if a contour is found

if largest_contour is None:

return image, None # No plate detected

# Approximate the contour to a rectangle

peri = cv2.arcLength(largest_contour, True)

approx = cv2.approxPolyDP(largest_contour, 0.02 * peri, True)

# Ensure the contour has 4 corners (rectangle)

if len(approx) != 4:

return image, None # Not a rectangular plate

# Extract corner points from the approximated polygon

corners = approx.reshape(4, 2)

# Sort the corners based on their y-coordinates

top_left, bottom_left, bottom_right, top_right = corners[corners[:, 1].argsort()]

# Calculate the width and height of the plate based on sorted corners

w = int(np.sqrt((bottom_right[0] - bottom_left[0])**2 + (bottom_right[1] - bottom_left[1])**2))

h = int(np.sqrt((top_right[0] - top_left[0])**2 + (top_right[1] - top_left[1])**2))

# Define the ROI (region of interest) for the cropped plate

x = int(bottom_left[0])

y = int(top_left[1])

cropped_plate = image[y:y+h, x:x+w]

return image, cropped_plate

# Load the car image

image = cv2.imread("carplate.jpg")

# Attempt to find and crop the license plate

original_image, cropped_plate = find_and_crop_plate(image.copy())

if cropped_plate is None:

print("No license plate detected.")

else:

# Display the original image and cropped plate

cv2.imshow("Original Image", original_image)

cv2.imshow("Cropped License Plate", cropped_plate)

cv2.waitKey(0)

cv2.destroyAllWindows()