ocr 图像倾斜矫正-radon变换

图像预处理

在ocr处理时候，可能遇到的图片会是倾斜的，导致检测不全问题，进而影响后续识别问题。

常见的倾斜矫正方法

• 霍夫变换轮廓检测
• randon 变换
• 基于PCA的方法

radon变换

本节说下randon变换

• 基本原理
  Radon（拉东）算法是一种通过定方向投影叠加，找到最大投影值时角度，从而确定图像倾斜角度的算法。具体过程如图所示
  

• 实现

import cv2
import sys
import time
import numpy as np

from skimage.transform import radon


filename = sys.argv[1]
# Load file, converting to grayscale
t1 = time.time()
img = cv2.imread(filename)
I = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
h, w = I.shape
# If the resolution is high, resize the image to reduce processing time.
# lower resolution, lower process time.
"""
if (w > 640):
    I = cv2.resize(I, (640, int((h / w) * 640)))
"""
if (w > 240):
    I = cv2.resize(I, (240, int((h / w) * 240)))

I = I - np.mean(I)  # Demean; make the brightness extend above and below zero
# Do the radon transform
sinogram = radon(I)
# Find the RMS value of each row and find "busiest" rotation,
# where the transform is lined up perfectly with the alternating dark
# text and white lines
r = np.array([np.sqrt(np.mean(np.abs(line) ** 2)) for line in sinogram.transpose()])
rotation = np.argmax(r)
print('Rotation: {:.2f} degrees'.format(90 - rotation))

# Rotate and save with the original resolution
M = cv2.getRotationMatrix2D((w/2, h/2), 90 - rotation, 1)
t2 = time.time()
print(t2-t1)
dst = cv2.warpAffine(img, M, (w, h))
cv2.imwrite('rotated.jpg', dst)
cv2.imshow('dst', dst)
cv2.waitKey()

结果

参考

• https://gist.github.com/endolith/334196bac1cac45a4893#file-rotation_spacing-py
• https://www.cnblogs.com/skyfsm/p/6902524.html
• https://blog.csdn.net/carson2005/article/details/40535199