Concat contours from image

Question

I am making plate recognition app. I managed to find contours of each element of plate. How can I cut them out, resize and combine, so tesseract ocr can do it's job and recognize the text from the plate. To make it more clear I attach pictures from the article.

Answer 1

This is pretty straight forward. Find the contours, loop through the contours, find bounding rect of each contour, crop out the part, resize the cropped part and store them in a list. You can then iterate the new list and join them using numpy's hstack. I've shared the full code, however you need some filtering technique to remove redundant contours so take care of that.

code:

import cv2
import numpy as np

img = cv2.imread('1.png',0)

image, contours, hierarchy = cv2.findContours(img.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

boxes = list()
cropped_imageList = list()

for c in contours:
    #if you want to store the original coordinates 
    x,y,w,h = cv2.boundingRect(c)
    boxes.append((x,y,w,h))
    #crop out the part of the image and resize
    crop_img = img[y:y+h, x:x+w]
    crop_img = cv2.resize(crop_img,(50,50))
    cropped_imageList.append(crop_img)

#join them back
imstack = cropped_imageList[0]

for imgs in cropped_imageList[1:]:
    imstack = np.hstack((imstack,imgs))

NOTE: This will randomly join the letters found, one way to solve this is to sort the boxes list using x,y and then crop instead.

Why are you doing all this when tesseract can directly detect the text? If you want better results, you should go for modern approached using deep learning.