cv.py

import cv2, numpy as np
from utils import *
import matplotlib.pylab as plt
import os
import uiautomator2 as u2


def cv_imread(file_path):
    cv_img = cv2.imdecode(np.fromfile(file_path,dtype=np.uint8),-1)
    return cv_img

class UIMatcher:

    @staticmethod
    def RotateClockWise90(img):
        trans_img = cv2.transpose(img)
        new_img = cv2.flip(trans_img, 0)
        return new_img

    @staticmethod
    def findpic(screen,template_paths = ['img/tiaoguo.jpg']):
            #返回相对坐标
        '''
        检测各种按钮(头像?)
        @return: 中心坐标lists, 对应的可信度list
        '''
        zhongxings = []
        max_vals = []
            # 增加判断screen方向
        if screen.shape[0]>screen.shape[1]:
            screen = UIMatcher.RotateClockWise90(screen)
        screen_show = screen.copy()
        # screen = cv2.cvtColor(screen, cv2.COLOR_BGR2GRAY)
        # plt.imshow(screen)
        # plt.show()
        for template_path in template_paths:
            template = cv_imread(template_path)
            # template = cv2.cvtColor(template, cv2.COLOR_BGR2GRAY)cv_imread
            h, w = template.shape[:2]  # rows->h, cols->w
            res = cv2.matchTemplate(screen, template, cv2.TM_CCOEFF_NORMED)
            min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res)
            x = (max_loc[0]+w//2)/screen.shape[1]
            y = (max_loc[1] + h // 2) / screen.shape[0]
            zhongxings.append([x,y])
            max_vals.append(max_val)
            if max_val>0.8:
                cv2.rectangle(screen_show, (int(max_loc[0]), int(max_loc[1])), (int(max_loc[0]+w),int(max_loc[1]+h)), (0, 0, 255), 2)
                cv2.putText(screen_show,str(round(max_val, 3))+os.path.basename(template_path),(int(max_loc[0]), int(max_loc[1])-2),cv2.FONT_HERSHEY_SIMPLEX,0.3, (0, 0, 255),1)
        # cv2.rectangle(screen, (0, 0), (10, 10), (0, 0, 255), 2)
        plt.cla()
        img4 = cv2.cvtColor(screen_show, cv2.COLOR_BGR2RGB)
        plt.imshow(img4)
        plt.pause(0.01)
            # if max_val>yuzhi:
            #     match_flag = 1
            # else:
            #     match_flag = 0
            # ax.hist(res.reshape(-1,1), 100, facecolor='b', alpha=0.5, label="rand_mat")
            # plt.show()
        return zhongxings,max_vals

    @staticmethod
    def find_gaoliang(screen):
        '''
        检测高亮位置(忽略了上板边,防止成就栏弹出遮挡)
        @return: 高亮中心相对坐标[x,y]
        '''
        if screen.shape[0] > screen.shape[1]:
            screen = UIMatcher.RotateClockWise90(screen)
        gray = cv2.cvtColor(screen, cv2.COLOR_RGB2GRAY)
        ret, binary = cv2.threshold(gray, 130, 255, cv2.THRESH_BINARY)
        index_1 = np.mean(np.argwhere(binary[63:,:] == 255), axis=0).astype(int)

        screen = cv2.cvtColor(binary, cv2.COLOR_GRAY2RGB)
        cv2.circle(screen, (index_1[1], index_1[0] + 63), 10, (255, 0, 0), -1)

        plt.cla()
        plt.imshow(screen)
        plt.pause(0.01)
        print(len(np.argwhere(binary == 255)),len(np.argwhere(binary == 0)))
        return index_1[1]/screen.shape[1], (index_1[0]+63)/screen.shape[0]


#
# d = u2.connect()
# screen = d.screenshot(format="opencv")
# # screen = cv_imread('test.jpg')
# UIMatcher.find_gaoliang(screen)
# plt.show()