1 1 # Author：Winter Liu is coming! 2 2 import cv2 as cv 3 3 import numpy as np 4 4 import pytesseract 5 5 6 6 7 7 # 预处理，高斯滤波（用处不大），4次开操作 8 8 # 过滤轮廓唯一 9 9 def contour_demo(img): 1010 gray = cv.cvtColor(img, cv.COLOR_BGR2GRAY) 1111 gray = cv.GaussianBlur(gray, (5, 5), 1) 1212 ref, thresh = cv.threshold(gray, 127, 255, cv.THRESH_BINARY) 1313 kernel = np.ones((9, 9), np.uint8) 1414 thresh = cv.morphologyEx(thresh, cv.MORPH_OPEN, kernel, iterations=4) 1515 contours, hierachy = cv.findContours(thresh, cv.RETR_EXTERNAL, cv.CHAIN_APPROX_SIMPLE) 1616 print(len(contours)) 1717 return contours 1818 1919 2020 def capture(img): 2121 contours = contour_demo(img) 2222 # 轮廓唯一，以后可以扩展 2323 contour = contours[0] 2424 # 求周长，可在后面的转换中使用周长和比例 2525 print(cv.arcLength(contour,True)) 2626 img_copy = img.copy() 2727 # 使用approxPolyDP，将轮廓转换为直线，22为精度（越高越低），TRUE为闭合 2828 approx = cv.approxPolyDP(contour, 22, True) 2929 # print(approx.shape) 3030 # print(approx) 3131 # cv.drawContours(img_copy, [approx], -1, (255, 0, 0), 15) 3232 n = [] 3333 # 生产四个角的坐标点 3434 for x, y in zip(approx[:, 0, 0], approx[:, 0, 1]): 3535 n.append((x, y)) 3636 p1 = np.array(n, dtype=np.float32) 3737 # 对应点 3838 p2 = np.array([(0, 0), (0, 1500), (1000, 1500), (1000, 0)], dtype=np.float32) 3939 M = cv.getPerspectiveTransform(p1, p2) # 变换矩阵 4040 # 使用透视变换 4141 result = cv.warpPerspective(img_copy, M, (0, 0)) 4242 # 重新截取 4343 result = result[:1501, :1001] 4444 cv.imwrite(r"C:\PycharmProjects\OpenCV\pic\ocr.png", result) 4545 return result 4646 4747 4848 # 图像识别代码，需要预先下载安装开源工具包 pytesseract，配置环境变量 4949 # pip install pytesseract 5050 # 修改“C:\Python\Python37\Lib\site-packages\pytesseract\pytesseract.py”中“cmd”为绝对路径 5151 def ocr_img(img): 5252 gray = cv.cvtColor(img, cv.COLOR_BGR2GRAY) 5353 # 图像清晰度越高结果越精确，时间更长 5454 text = pytesseract.image_to_string(gray) 5555 print(text) 5656 5757 5858 src = cv.imread(r"C:\PycharmProjects\OpenCV\pic\page.jpg") 5959 res = capture(src) 6060 ocr_img(res) 6161 cv.waitKey(0) 6262 cv.destroyAllWindows()
