OpenCV旋转图像

K = 10
resized = cv2.resize(copied, (rows // K, cols // K))
domain = connected_domain(resized, 20)  # 缩小图像，以便加速图像分割
for label in domain:
    if (area := len(domain[label])) < 1440000 // (K * K):
        continue  # 舍弃掉面积小的连通域

    # 找 连通域的左上角和右下角
    x1, y1, x3, y3 = find_corners(domain[label])
    row = int(x1 / (1330.0 / K))  # 1330 为die 重复间距
    col = int(y1 / (1330.0 / K))
    index = 3 * row + col

    #print(f"row={row}, col={col}")
    die_bin = MX_MY_BIN_list[index][2]
    if die_bin == "GOOD":  # only analyze good die
        # print(f"row={row}, col={col}")
        # print(MX_MY_BIN_list[index], x1, y1)

        die_image = img[x1 * K:x3 * K, y1 * K:y3 * K]
        # plt.imshow(die_image, cmap="gray")
        # plt.title("-".join(MX_MY_BIN_list[index]))
        # plt.show()
        row1, col1, row2, col2 = find_left_dummy_pads(die_image)

        # 计算旋转角度
        theta = 0
        if (delta_y := col2 - col1) != 0:  # 象鼻表达式的优先级很低，所有这里需要括号
            theta = atan((row2 - row1) / delta_y) / pi * 180
        #print(f" die 的旋转角度 theta = {theta} 度")
        angle = (theta - 90) if theta > 0 else (theta + 90)
        M = cv2.getRotationMatrix2D((0, 0), angle, 1)  # 旋转中心，角度degree，放大比例
        die_image = cv2.warpAffine(die_image, M, die_image.shape)

        # plt.imshow(die_image, cmap="gray")
        # plt.title("rotated die image")
        # plt.show()