发票数据识别

最新推荐文章于 2024-07-30 11:52:42 发布

weixin_38092754

最新推荐文章于 2024-07-30 11:52:42 发布

阅读量2.8k

点赞数

本文链接：https://blog.csdn.net/weixin_38092754/article/details/81177043

版权

该博客介绍了使用OpenCV和PyTesseract进行图像处理，通过膨胀、腐蚀等操作筛选发票上的文字区域，然后通过轮廓分析和阈值处理进一步提取文字，最终实现发票文字的识别与定位。

摘要由CSDN通过智能技术生成

import cv2
import math
import numpy as np
import matplotlib.pyplot as plt
import skimage
from PIL import Image
from pytesseract import pytesseract
from skimage import data,color,morphology,feature
import argparse
#import cvHelper

# 原始图像
img_ori1 = cv2.imread('TestData/taxi/IMG_3787.JPG')
img_ori2 = cv2.imread('TestData/taxi/IMG_3789.JPG')


imgs = [img_ori2]
resize_imgs = []
gray_resize_imgs = []

# 缩放图像
for idx,im in enumerate(imgs):
    width = 300.0          # 缩放 目标宽度
    r = width/im.shape[1] # 缩放因子
    dim = (int(width), int(im.shape[0]*r))
    img_resized = cv2.resize(im, dim, interpolation=cv2.INTER_AREA)
    resize_imgs.append(img_resized)
    gray = cv2.cvtColor(img_resized, cv2.COLOR_BGR2GRAY)
    gray_resize_imgs.append(gray)

# 显示图像
import pylab
cv2.namedWindow("ori img", cv2.WINDOW_AUTOSIZE)
cv2.moveWindow('ori img', 20, 24)
cv2.imshow('ori img', resize_imgs[0])
pylab.show()

im_at_mean = cv2.adaptiveThreshold(gray_resize_imgs[0], 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY_INV, 3, 5)
cv2.imshow("im_at_mean", im_at_mean)
pylab.show()

b,g,r = cv2.split(resize_imgs[0])
th, dst = cv2.threshold(r, 160, 255, cv2.THRESH_BINARY)
cv2.imshow("r_threshold", dst)
pylab.show()

# 膨胀
kernel = np.ones((3, 3), np.uin