11.4OpenCV_图像预处理习题02

1.身份证号码识别（结果：身份证号识别结果为：911124198108030024）

import cv2
import numpy as np
import paddlehub as hub

def get_text():
    img = cv2.imread("images1/images/shenfen03.jpg")
    # 灰度化
    gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    # 高斯
    gs_img = cv2.GaussianBlur(gray_img, (9, 9), 0)
    # 腐蚀
    ero_img = cv2.erode(gs_img, np.ones((11, 11), np.uint8))
    # 边缘
    cany_img = cv2.Canny(ero_img, 70, 300)
    cv2.imshow("Canny Image", cany_img)
    # 轮廓
    contours, _ = cv2.findContours(cany_img, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
    # 创建一个与原始图像同样大小的黑色图像
    contour_img = np.zeros_like(img)
    # 在黑色图像上绘制轮廓
    cv2.drawContours(contour_img, contours, -1, (255, 255, 255), 2)
    # 显示轮廓图像
    cv2.imshow("Contours", contour_img)

    for contour in contours:
        x, y, w, h = cv2.boundingRect(contour)
        print(w,h)
        if w > 200 and h < 70:
            cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
            out_img = img[y:y + h, x:x + w]

    # # 绘制所有轮廓的矩形框
    # for contour in contours:
    #     x, y, w, h = cv2.boundingRect(contour)
    #     # 移除条件判断，为每个轮廓绘制矩形框
    #     cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
    cv2.imshow("title", img)
    # 显示原始图像上的矩形
    cv2.imshow("title", out_img)
    cv2.waitKey(0)


    #加载模型
    ocr = hub.Module(name="chinese_ocr_db_crnn_server")
#识别文本
    results = ocr.recognize_text(images=[out_img])
    for result in results:
        data = result['data']
        for x in data:
            print('文本: ', x['text'])

    cv2.destroyAllWindows()

if __name__ == "__main__":
    get_text()

2.车牌识别

import cv2
import numpy as np
import paddlehub as hub

def get_text():
    img = cv2.imread("images1/images/car6.png")
    # 灰度化
    gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    # 顶帽
    eroded = cv2.morphologyEx(gray_img, cv2.MORPH_TOPHAT, np.ones((9,9), np.uint8))
    # 高斯
    gs_img = cv2.GaussianBlur(eroded, (9, 9), 2)
    # 边缘
    cany_img = cv2.Canny(gs_img, 170, 180)
    # 膨胀
    eroded2 = cv2.dilate(cany_img,np.ones((17,17), np.uint8), iterations=2)
    # 轮廓
    contours, _ = cv2.findContours(eroded2, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
    # 创建一个与原始图像同样大小的黑色图像
    contour_img = np.zeros_like(img)
    # 在黑色图像上绘制轮廓
    cv2.drawContours(contour_img, contours, -1, (255, 255, 255), 2)
    # 显示轮廓图像
    cv2.imshow("Contours", contour_img)

    for contour in contours:
        x, y, w, h = cv2.boundingRect(contour)
        print(w,h)
        if w > 20 and h > 20:
            cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
            out_img = img[y:y + h, x:x + w]
        cv2.imshow("title", out_img)
        cv2.waitKey(0)
        cv2.destroyAllWindows()
    # # 绘制所有轮廓的矩形框
    # for contour in contours:
    #     x, y, w, h = cv2.boundingRect(contour)
    #     # 移除条件判断，为每个轮廓绘制矩形框
    #     cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
    # cv2.imshow("title", img)
    # 显示原始图像上的矩形

    #加载模型
    ocr = hub.Module(name="chinese_ocr_db_crnn_server")
#识别文本
    results = ocr.recognize_text(images=[out_img])
    for result in results:
        data = result['data']
        for x in data:
            print('文本: ', x['text'])



if __name__ == "__main__":
    get_text()

原文地址：https://blog.csdn.net/gs1we1/article/details/143495307

免责声明：本站文章内容转载自网络资源，如本站内容侵犯了原著者的合法权益，可联系本站删除。更多内容请关注自学内容网（zxcms.com）！