基于Python和OpenCV实现实时文档扫描的全流程

pip install opencv-python numpy 

import numpy as np
import cv2
 
def order_points(pts):
    # 初始化4个顶点的坐标（左上、右上、右下、左下）
    rect = np.zeros((4, 2), dtype="float32")
    
    # 1. 计算每个点的x+y之和：左上点的和最小，右下点的和最大
    s = pts.sum(axis=1)
    rect[0] = pts[np.argmin(s)]  # 左上（tl）
    rect[2] = pts[np.argmax(s)]  # 右下（br）
    
    # 2. 计算每个点的y-x之差（np.diff默认是后减前，即y-x）：右上点的差最小，左下点的差最大
    diff = np.diff(pts, axis=1)
    rect[1] = pts[np.argmin(diff)]  # 右上（tr）
    rect[3] = pts[np.argmax(diff)]  # 左下（bl）
    
    return rect

def four_point_transform(image, pts):
    # 1. 获取排序后的四个顶点
    rect = order_points(pts)
    (tl, tr, br, bl) = rect  # 左上、右上、右下、左下
    
    # 2. 计算目标矩形的宽度（取文档左右两边的最大长度，避免变形）
    widthA = np.sqrt(((br[0] - bl[0]) ** 2) + ((br[1] - bl[1]) ** 2))  # 下边长
    widthB = np.sqrt(((tr[0] - tl[0]) ** 2) + ((tr[1] - tl[1]) ** 2))  # 上边长
    maxWidth = max(int(widthA), int(widthB))  # 目标宽度
    
    # 3. 计算目标矩形的高度（取文档上下两边的最大长度）
    heightA = np.sqrt(((tr[0] - br[0]) ** 2) + ((tr[1] - br[1]) ** 2))  # 右边长
    heightB = np.sqrt(((tl[0] - bl[0]) ** 2) + ((tl[1] - bl[1]) ** 2))  # 左边长
    maxHeight = max(int(heightA), int(heightB))  # 目标高度
    
    # 4. 定义目标矩形的四个顶点
    dst = np.array([
        [0, 0],                  # 左上
        [maxWidth - 1, 0],       # 右上（减1是因为像素索引从0开始）
        [maxWidth - 1, maxHeight - 1],  # 右下
        [0, maxHeight - 1]], dtype="float32")  # 左下
    
    # 5. 计算透视矩阵M
    M = cv2.getPerspectiveTransform(rect, dst)
    
    # 6. 应用透视变换，得到矫正后的图像
    warped = cv2.warpPerspective(image, M, (maxWidth, maxHeight))
    
    return warped

# 图像预处理
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)  # 灰度化
gray = cv2.GaussianBlur(gray, (5, 5), 0)  # 高斯模糊（5x5卷积核，标准差0）
edged = cv2.Canny(gray, 15, 45)  # 边缘检测（低阈值15，高阈值45）
 
# 轮廓检测与筛选
cnts = cv2.findContours(edged, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)[-2]  # 提取外部轮廓
cnts = sorted(cnts, key=cv2.contourArea, reverse=True)[:3]  # 按面积降序，取前3个
 
# 遍历轮廓，判断是否为文档（四边形+面积足够大）
for c in cnts:
    peri = cv2.arcLength(c, True)  # 计算轮廓周长（True表示轮廓闭合）
    # 多边形逼近：epsilon=0.05*peri（逼近精度，值越小越接近原轮廓）
    approx = cv2.approxPolyDP(c, 0.05 * peri, True)
    area = cv2.contourArea(approx)  # 计算逼近后多边形的面积
    
    # 条件：面积>20000（过滤小物体）且顶点数=4（文档为四边形）
    if area > 20000 and len(approx) == 4:
        screenCnt = approx  # 确定文档轮廓
        print("检测到文档，轮廓面积：", area)
        break

# 对透视变换后的图像进行二值化
warped_gray = cv2.cvtColor(warped_result, cv2.COLOR_BGR2GRAY)
# 自动阈值二值化（THRESH_OTSU：自动计算最优阈值，避免手动调参）
ref_result = cv2.threshold(warped_gray, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)[1]

# 自定义显示函数（不自动关闭窗口，需按q退出）
def cv_show(name, img):
    cv2.imshow(name, img)
 
# 初始化摄像头（0表示默认摄像头）
cap = cv2.VideoCapture(0)
 
# 检查摄像头是否正常打开
if not cap.isOpened():
    print("无法打开摄像头")
    exit()
 
# 实时采集与处理循环
while True:
    ret, image = cap.read()  # 读取一帧图像（ret：是否读取成功，image：图像数据）
    orig = image.copy()  # 保存原始图像副本
    
    if not ret:
        print("无法读取摄像头帧")
        break
    
    # 1. 显示原始图像
    cv_show("Original", image)
    
    # 2. 图像预处理与轮廓检测（此处省略，见模块3）
    # ...（预处理、轮廓检测代码）...
    
    # 3. 若检测到文档，显示结果
    if flag == 1:
        # 绘制文档轮廓
        image_with_doc = cv2.drawContours(orig.copy(), [screenCnt], 0, (0, 255, 0), 2)
        cv_show("Document Detection", image_with_doc)
        
        # 透视变换
        warped_result = four_point_transform(orig, screenCnt.reshape(4, 2))
        cv_show("Warped", warped_result)
        
        # 二值化
        ref_result = cv2.threshold(cv2.cvtColor(warped_result, cv2.COLOR_BGR2GRAY), 
                                   0, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)[1]
        cv_show("Binarized", ref_result)
    
    # 按下'q'键退出循环（waitKey(1)：等待1ms，返回按键ASCII码）
    if cv2.waitKey(1) == ord('q'):
        break
 
# 释放摄像头资源，关闭所有窗口
cap.release()
cv2.destroyAllWindows()

import numpy as np
import cv2
 
 
def order_points(pts):
    rect = np.zeros((4, 2), dtype="float32")
    s = pts.sum(axis=1)
    rect[0] = pts[np.argmin(s)]
    rect[2] = pts[np.argmax(s)]
    diff = np.diff(pts, axis=1)
    rect[1] = pts[np.argmin(diff)]
    rect[3] = pts[np.argmax(diff)]
    return rect
 
 
def four_point_transform(image, pts):
    rect = order_points(pts)
    (tl, tr, br, bl) = rect
    widthA = np.sqrt(((br[0] - bl[0]) ** 2) + ((br[1] - bl[1]) ** 2))
    widthB = np.sqrt(((tr[0] - tl[0]) ** 2) + ((tr[1] - tl[1]) ** 2))
    maxWidth = max(int(widthA), int(widthB))
    heightA = np.sqrt(((tr[0] - br[0]) ** 2) + ((tr[1] - br[1]) ** 2))
    heightB = np.sqrt(((tl[0] - bl[0]) ** 2) + ((tl[1] - bl[1]) ** 2))
    maxHeight = max(int(heightA), int(heightB))
    dst = np.array([
        [0, 0],
        [maxWidth - 1, 0],
        [maxWidth - 1, maxHeight - 1],
        [0, maxHeight - 1]], dtype="float32")
    M = cv2.getPerspectiveTransform(rect, dst)
    warped = cv2.warpPerspective(image, M, (maxWidth, maxHeight))
    return warped
 
 
def cv_show(name, img):
    cv2.imshow(name, img)
 
 
if __name__ == "__main__":
    cap = cv2.VideoCapture(0)
    if not cap.isOpened():
        print("Cannot open camera")
        exit()
 
    while True:
        flag = 0
        ret, image = cap.read()
        orig = image.copy()
        if not ret:
            print("不能读取摄像头")
            break
 
        # 显示原始图像
        cv_show("Original", image)
 
        # 图像预处理
        gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
        gray = cv2.GaussianBlur(gray, (5, 5), 0)
        edged = cv2.Canny(gray, 15, 45)
        cv_show("Edge Detection", edged)
 
        # 轮廓检测与筛选
        cnts = cv2.findContours(edged, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)[-2]
        cnts = sorted(cnts, key=cv2.contourArea, reverse=True)[:3]
        image_contours = cv2.drawContours(image.copy(), cnts, -1, (0, 255, 0), 2)
        cv_show("Contours", image_contours)
 
        # 识别文档轮廓
        for c in cnts:
            peri = cv2.arcLength(c, True)
            approx = cv2.approxPolyDP(c, 0.05 * peri, True)
            area = cv2.contourArea(approx)
 
            if area > 20000 and len(approx) == 4:
                screenCnt = approx
                flag = 1
                print(f"检测到文档，周长：{peri:.2f}，面积：{area:.2f}")
 
                # 显示文档检测结果
                image_with_doc = cv2.drawContours(orig.copy(), [screenCnt], 0, (0, 255, 0), 2)
                cv_show("Document Detection", image_with_doc)
 
                # 透视变换
                warped_result = four_point_transform(orig, screenCnt.reshape(4, 2))
                cv_show("Warped", warped_result)
 
                # 二值化
                warped_gray = cv2.cvtColor(warped_result, cv2.COLOR_BGR2GRAY)
                ref_result = cv2