# -*- coding: utf-8 -*-
import copy
import time
import cv2
import os
import numpy as np
def get_view(small_img,radius=70):
    """
    返回当前视野角度 感谢网友飞宇提供的识别方法
    Returns:以正北方向为0度计算

    """

    # radius = 70  # 定义截屏的宽度和长度
    # # 获取窗口句柄
    # hwnd = win32gui.FindWindow("UnityWndClass", "原神")  # 替换成你实际的窗口句柄
    #
    #
    # small_img = screenshot(hwnd,(98+radius,55+radius), radius=radius)
    #
    #
    channels = cv2.split(small_img)
    # 提取Alpha通道
    alpha3 = channels[3]  # 索引为3，因为Alpha通道是第四个通道
    # cv2.imshow('Alpha Channel', alpha3)
    # cv2.waitKey(0)
    # cv2.destroyAllWindows()
    # _, alpha = cv2.threshold(small_img, 10, 255, cv2.THRESH_BINARY) #这个是提取蓝色箭头图像
    # 假设 gray_channel 是你的灰度图像
    # 创建一个掩膜，将150-200之间的值设为255，其他设为0
    mask = np.zeros_like(alpha3)
    mask[alpha3 >= 151] = 255
    mask[alpha3 < 151] = 0
    mask[alpha3 > 229] = 0
    img = np.zeros((radius * 2, radius * 2), dtype=np.uint8)
    center = (radius, radius)
    max_overlap = 0  # 最佳的重叠面积
    max_angle = -1  # 最佳的角度
    increment = 20  # 递增的角度，这里设置为10度
    for angle in range(0, 360, increment):
        img_copy = img.copy()
        # 画一个扇形
        cv2.ellipse(img_copy, center, (radius, radius), 225 + angle, 0, 90, 255, -1)
        overlap = np.logical_and(img_copy, mask).sum()
        if overlap > max_overlap:
            max_overlap = overlap
            max_angle = angle

    start_angle = max_angle - 10
    if start_angle < 0:
        start_angle = 0
        end_angle = start_angle + 10
    else:
        end_angle = start_angle + 20

    for angle in range(start_angle, end_angle, 1):
        img_copy = img.copy()
        # 画一个扇形
        cv2.ellipse(img_copy, center, (radius, radius), 225 + angle, 0, 90, 255, -1)
        overlap = np.logical_and(img_copy, mask).sum()
        if overlap > max_overlap:
            max_overlap = overlap
            max_angle = angle
    # print(f'最匹配角度:{max_angle}，重叠面积:{max_overlap}')
    # print(f'用时：{time.time() - old_time}')
    #cv2.destroyAllWindows()

    return max_angle
def get_view2(small_img,radius=70):
    """
    返回当前视野角度 感谢网友飞宇提供的识别方法
    Returns:以正北方向为0度计算

    """

    # radius = 70  # 定义截屏的宽度和长度
    # # 获取窗口句柄
    # hwnd = win32gui.FindWindow("UnityWndClass", "原神")  # 替换成你实际的窗口句柄
    #
    #
    # small_img = screenshot(hwnd,(98+radius,55+radius), radius=radius)
    #
    #
    channels = cv2.split(small_img)
    # 提取Alpha通道
    alpha3 = channels[3]  # 索引为3，因为Alpha通道是第四个通道
    #cv2.imshow('Alpha Channel', alpha3)
    #cv2.waitKey(0)
    #cv2.destroyAllWindows()
    # _, alpha = cv2.threshold(small_img, 10, 255, cv2.THRESH_BINARY) #这个是提取蓝色箭头图像
    # 假设 gray_channel 是你的灰度图像
    # 创建一个掩膜，将150-200之间的值设为255，其他设为0
    mask = np.zeros_like(alpha3)
    mask[alpha3 >= 9] = 255
    mask[alpha3 < 9] = 0
    mask[alpha3 > 190] = 0
    img = np.zeros((radius * 2, radius * 2), dtype=np.uint8)
    center = (radius, radius)
    max_overlap = 0  # 最佳的重叠面积
    max_angle = -1  # 最佳的角度
    increment = 20  # 递增的角度，这里设置为10度
    for angle in range(0, 360, increment):
        img_copy = img.copy()
        # 画一个扇形
        cv2.ellipse(img_copy, center, (radius, radius), 225 + angle, 0, 90, 255, -1)
        overlap = np.logical_and(img_copy, mask).sum()
        if overlap > max_overlap:
            max_overlap = overlap
            max_angle = angle

    start_angle = max_angle - 10
    if start_angle < 0:
        start_angle = 0
        end_angle = start_angle + 10
    else:
        end_angle = start_angle + 20

    for angle in range(start_angle, end_angle, 1):
        img_copy = img.copy()
        # 画一个扇形
        cv2.ellipse(img_copy, center, (radius, radius), 225 + angle, 0, 90, 255, -1)
        overlap = np.logical_and(img_copy, mask).sum()
        if overlap > max_overlap:
            max_overlap = overlap
            max_angle = angle


    return max_angle
def pyramid_template_matching(image, template, mask=None):
    '''
    返回当前导航的角度
    :param image:
    :param template:
    :param mask:
    :return:
    '''
    result = image.copy()
    # 创建掩码
    if mask is not None:
        mask = mask.astype(np.uint8)
    # 金字塔层级
    #pyramid_levels =1
    best_angle = 0  # 最佳匹配角度
    max_similarity = 0  # 最大相似度
    best_res=0
    best_rotated_template=""
    h, w = template.shape[:2]
    scaled_mask = mask.astype(np.uint8)
    scaled_image=image
    scaled_template=template
    # for level in range(pyramid_levels):
    #     # 缩放大图和小图
    #     scaled_image = cv2.resize(image, (0, 0), fx=1 / (2 ** level), fy=1 / (2 ** level))
    #     scaled_template = cv2.resize(template, (0, 0), fx=1 / (2 ** level), fy=1 / (2 ** level))
    #     if mask is not None:
    #         scaled_mask = cv2.resize(mask, (0, 0), fx=1 / (2 ** level), fy=1 / (2 ** level))
    #         scaled_mask = scaled_mask.astype(np.uint8)
    # 旋转模板并进行匹配
    for angle in range(0, 360, 10):
        rotated_mask = rotate_image(scaled_mask, angle)
        rotated_mask[rotated_mask >1] = 255 # 将灰色也变成白色
        rotated_template = rotate_image(scaled_template, angle)
        #save_rotated_image(rotated_template, "./datas/img", f"{angle}_template")
        #save_rotated_image(rotated_mask, "./datas/img",f"{angle}_mask")
        res = cv2.matchTemplate(scaled_image, rotated_template, cv2.TM_CCORR_NORMED, mask=rotated_mask)
        similarity = cv2.minMaxLoc(res)[1]  # 获取匹配相似度
        if similarity > max_similarity:
            max_similarity = similarity
            best_angle = angle
            best_res= res
            best_rotated_template = rotated_template

    best_angle_bk= copy.copy(best_angle)
    start_angle=best_angle_bk-10
    if start_angle<0:
        start_angle=0
        end_angle=start_angle+10
    else:
        end_angle = start_angle + 20
    for angle in range(start_angle,end_angle,1):
        rotated_mask = rotate_image(scaled_mask, angle)
        rotated_mask[rotated_mask > 1] = 255  # 将灰色像素值改为黑色像素值
        rotated_template = rotate_image(scaled_template, angle)
        res = cv2.matchTemplate(scaled_image, rotated_template, cv2.TM_CCORR_NORMED, mask=rotated_mask)
        similarity = cv2.minMaxLoc(res)[1]  # 获取匹配相似度
        if similarity > max_similarity:
            max_similarity = similarity
            best_angle = angle
            best_res = res
            best_rotated_template=rotated_template

    #min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(best_res)
    #top_left = max_loc
    #bottom_right = (top_left[0] + w, top_left[1] + h)
    # 绘制角度
    #cv2.putText(result, f"{int(best_angle)}", (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)
    # 在原始大图上绘制矩形框标记匹配区域
    #cv2.rectangle(result, top_left, bottom_right, (0, 255, 0), 2)
    return result,max_similarity,best_angle,best_rotated_template
def save_rotated_image(rotated_image, save_dir, index):
    save_path = os.path.join(save_dir, f"{index}.png")
    cv2.imwrite(save_path, rotated_image)
def rotate_image(image, angle):
    '''
    将图片旋转多少度
    :param image:
    :param angle:
    :return:
    '''
    h, w = image.shape[:2]
    center = (w // 2, h // 2)
    M = cv2.getRotationMatrix2D(center, -angle, 1.0)
    rotated_image = cv2.warpAffine(image, M, (w, h), flags=cv2.INTER_NEAREST, borderMode=cv2.BORDER_CONSTANT, borderValue=(0, 0, 0, 0))
    return rotated_image
def template_matching(image, template, threshold=0.8,maxcnt=0,mask=None):
    '''
    模板匹配 多个
    :param image:
    :param template:
    :param mask:
    :return:
    '''
    method = cv2.TM_CCOEFF_NORMED
    # 创建掩码
    if mask is not None:
        mask = mask.astype(np.uint8)
        scaled_mask = mask.astype(np.uint8)
    else:
        scaled_mask=mask
    scaled_image=image
    scaled_template=template
    h, w = scaled_template.shape[:2]
    res = cv2.matchTemplate(scaled_image, scaled_template, method, mask=scaled_mask)

    result = []
    while True:
        min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res)
        if method in [cv2.TM_SQDIFF, cv2.TM_SQDIFF_NORMED]:
            top_left = min_loc
        else:
            top_left = max_loc
        if max_val < threshold or max_val == np.inf:
            break
        middle_point = [top_left[0] + w // 2, top_left[1] + h // 2]
        result.append(dict(
            result=middle_point,
            rectangle=(top_left, (top_left[0], top_left[1] + h), (top_left[0] + w, top_left[1]),
                       (top_left[0] + w, top_left[1] + h)),
            confidence=round(max_val, 2)
        ))

        if maxcnt and len(result) >= maxcnt:
            break
        cv2.floodFill(res, None, max_loc, (-1000,), max_val - threshold + 0.1, 1, flags=cv2.FLOODFILL_FIXED_RANGE)
    return result

if __name__ == '__main__':
    # 读取图像和模板
    # old_time=time.time()
    # image = cv2.imdecode(np.fromfile(file=r"大图.png", dtype=np.uint8), cv2.IMREAD_UNCHANGED)  # 加载大图
    # template = cv2.imdecode(np.fromfile(file=r"小图.png", dtype=np.uint8), cv2.IMREAD_UNCHANGED)  # 加载透明图
    # mask = template[:, :, 3]  # 提取透明度通道作为掩码
    # result,max_similarity, best_angle = pyramid_template_matching(image, template, mask=mask)
    # print(f"相似度:{max_similarity} 角度:{best_angle}  用时:{int((time.time()-old_time)*1000)}ms")
    #
    # # 显示结果图像
    # cv2.imshow("Result", result)
    # cv2.waitKey(0)
    # cv2.destroyAllWindows()

    old_time=time.time()
    image = cv2.imdecode(np.fromfile(file=r"D:\QQ聊天记录\111.bmp", dtype=np.uint8), cv2.IMREAD_COLOR)  # 加载大图
    template = cv2.imdecode(np.fromfile(file=r"D:\QQ聊天记录\p.png", dtype=np.uint8), cv2.IMREAD_COLOR)  # 加载透明图

    RET = template_matching(image, template, mask=None)
    for result in RET:
        print(result)
        # 转换为 NumPy 数组
        pts = np.array((result['rectangle'][0], result['rectangle'][1], result['rectangle'][3],
                        result['rectangle'][2]), np.int32)



        cv2.polylines(image, [pts], True, (0, 255, 0), 5)


        # 显示结果图像
        cv2.imshow("324234", result)
        cv2.waitKey(0)
        cv2.destroyAllWindows()