二维码的图像特征

import cv2
import numpy as np
import math

def preprocess_image(img):
    """预处理：灰度化、二值化、去噪"""
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    blur = cv2.GaussianBlur(gray, (5, 5), 0)
    # 二值化（黑底白字转白底黑字，便于分析）
    _, thresh = cv2.threshold(blur, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)
    # 形态学操作去除小噪点
    kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (3, 3))
    thresh = cv2.morphologyEx(thresh, cv2.MORPH_CLOSE, kernel)
    return thresh

def is_square(contour, eps=0.05):
    """判断轮廓是否为正方形（边长比接近1，且为凸多边形）"""
    perimeter = cv2.arcLength(contour, True)
    if perimeter == 0:
        return False
    approx = cv2.approxPolyDP(contour, eps * perimeter, True)
    # 正方形需4个顶点，且凸性良好，宽高比接近1
    return (len(approx) == 4 and 
            cv2.isContourConvex(approx) and 
            0.9 <= (cv2.boundingRect(approx)[2] / cv2.boundingRect(approx)[3]) <= 1.1)

def check_position_pattern(roi):
    """验证区域是否为二维码定位点（回字形结构）"""
    h, w = roi.shape[:2]
    if h != w:  # 必须为正方形
        return False
    
    # 缩放为固定尺寸（便于比例计算）
    size = 20  # 缩小到20x20像素
    roi_resized = cv2.resize(roi, (size, size))
    _, roi_bin = cv2.threshold(roi_resized, 127, 255, cv2.THRESH_BINARY)  # 确保二值化
    
    # 定位点结构：外层黑框（占比~70%）、中层白框（占比~50%）、中心黑方块（占比~30%）
    outer_ratio = np.sum(roi_bin == 0) / (size * size)  # 黑色像素占比（外层）
    if not (0.6 <= outer_ratio <= 0.8):  # 外层黑框占比约60%-80%
        return False
    
    # 裁剪中间区域（去掉外层10%边框）
    inner_size = int(size * 0.8)
    inner_roi = roi_bin[int(size*0.1):int(size*0.9), int(size*0.1):int(size*0.9)]
    inner_white_ratio = np.sum(inner_roi == 255) / (inner_size * inner_size)  # 中层白框占比
    if not (0.4 <= inner_white_ratio <= 0.6):  # 中层白框占比约40%-60%
        return False
    
    # 裁剪中心区域（去掉中层后剩余30%）
    center_size = int(size * 0.3)
    center_roi = roi_bin[int(size*0.35):int(size*0.65), int(size*0.35):int(size*0.65)]
    center_black_ratio = np.sum(center_roi == 0) / (center_size * center_size)  # 中心黑方块占比
    return 0.7 <= center_black_ratio <= 1.0  # 中心几乎全黑

def check_geometry(points):
    """验证三个定位点的几何分布（近似直角三角形）"""
    if len(points) != 3:
        return False
    
    # 计算三点之间的距离
    d1 = math.hypot(points[0][0] - points[1][0], points[0][1] - points[1][1])
    d2 = math.hypot(points[1][0] - points[2][0], points[1][1] - points[2][1])
    d3 = math.hypot(points[0][0] - points[2][0], points[0][1] - points[2][1])
    dists = sorted([d1, d2, d3])
    
    # 验证勾股定理（直角三角形）：a² + b² ≈ c²（误差允许15%）
    a, b, c = dists
    if not (0.85 * c**2 <= a**2 + b**2 <= 1.15 * c**2):
        return False
    
    # 验证最长边与最短边比例（二维码定位点距离比例通常在1.5-3之间）
    if not (1.5 <= c / a <= 3.0):
        return False
    
    return True

def is_qrcode_region(img):
    """判断输入图块是否为二维码区域"""
    if img is None or img.size == 0:
        return False, []
    
    thresh = preprocess_image(img)
    contours, _ = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    
    position_points = []
    # 筛选候选定位点
    for cnt in contours:
        if is_square(cnt):
            x, y, w, h = cv2.boundingRect(cnt)
            roi = thresh[y:y+h, x:x+w]  # 裁剪候选区域
            if check_position_pattern(roi):
                # 记录定位点中心坐标
                center = (x + w//2, y + h//2)
                position_points.append(center)
                # 绘制候选定位点（调试用）
                cv2.rectangle(img, (x, y), (x+w, y+h), (0, 255, 0), 2)
    
    # 必须检测到3个定位点，且几何分布符合要求
    if len(position_points) == 3 and check_geometry(position_points):
        return True, position_points
    else:
        return False, position_points

# 测试代码
if __name__ == "__main__":
    img = cv2.imread("qrcode_test.jpg")  # 输入待检测的图块
    is_qr, points = is_qrcode_region(img)
    
    if is_qr:
        print("该图块是二维码区域")
        # 绘制定位点中心
        for (x, y) in points:
            cv2.circle(img, (x, y), 5, (0, 0, 255), -1)
    else:
        print("该图块不是二维码区域")
    
    cv2.imshow("Result", img)
    cv2.waitKey(0)
    cv2.destroyAllWindows()