import matplotlib.pyplot as plt
from PIL import Image
import adbOperate
import image_analysis
import time
import random
import cv2class ImageHelprt:#定义小人坐标peo_loc=(0,0)#定义跳跃目标的坐标tar_loc=(0,0)#定义鼠标点击类型click_count=1reg_event=Truedef __init__(self):self.figure=plt.figure()self.tempPath=''self.count=0self.img_analysis=image_analysis.IMG_Aanlysis()self.CLICK_SOURCE=(500,1514)self.exit=False#设置限制跳跃次数self.limit_count=50def OnBtn_pressEvent(self, event):'''鼠标点击图片时触发的时事件:param event::return:'''eX=event.xdataeY=event.ydataif(self.click_count==0):self.click_count=1else:print('点击位置 X:', eX, '   Y:', eY)#第二次点击计算距离self.tar_loc=(eX,eY)self.click_count=1distance=self.cal_distance(self.peo_loc,self.tar_loc)click_loc=self.cal_click_loc(self.CLICK_SOURCE)#操作ADB 控制手机跳跃adbOperate.jump(distance,click_loc=click_loc)self.draw_result(self.cv_img, self.peo_loc, (0,0),(int(self.tar_loc[0]),int(self.tar_loc[1])) , click_loc)plt.close()def show_img(self):'''显示照片到当前figure中:param img: 要显示的图片:return:'''#读取截图self.img= Image.open(self.tempPath)self.cv_img=cv2.imread(self.tempPath)if(self.jump_count>=self.limit_count):self.exit=True#清空GDI容器self.figure.clear()#注册鼠标点击事件if(self.reg_event):self.figure.canvas.mpl_connect('button_press_event', self.OnBtn_pressEvent)#分析图片#获取小人位置(cv_img,peo_loc)=self.img_analysis.get_people_loc(self.cv_img)# 获取目标位置(cv_img,target_loc)=self.img_analysis.get_target_loc(self.cv_img)if(target_loc[0]==target_loc[1]==0):self.figure=plt.figure(num='请用鼠标点击目标棋子中心')#说明没有识别到下一跳中心位置，弹出手动选择self.figure.canvas.mpl_connect('button_press_event', self.OnBtn_pressEvent)print('请手动选择目标！')self.peo_loc=peo_locplt.imshow(self.cv_img,animated= True)plt.show()print('分析中...')else:#自动跳跃#计算距离distance=self.cal_distance(peo_loc=peo_loc,tar_loc=target_loc)#print('距离：%d' %distance)click_loc = self.cal_click_loc(self.CLICK_SOURCE)adbOperate.jump(distance,click_loc)self.draw_result(cv_img,peo_loc,target_loc,(0,0),click_loc)self.pull_screenshot()def cal_distance(self,peo_loc,tar_loc):'''计算小人与目标方块中心的最短距离:param peo_loc:小人位置坐标:param tar_loc:目标方块位置坐标:return:计算结果'''distance = (tar_loc[0] - peo_loc[0]) ** 2 + (tar_loc[1] - peo_loc[1]) ** 2distance = distance ** 0.5return distancedef cal_click_loc(self,source_loc):'''随机产生一个点击位置:param source_loc:源坐标:return: 源坐标+随机数'''click_x=500+random.randint(1,30)click_y=1514+random.randint(1,30)return (int(click_x),int(click_y))def pull_screenshot(self):'''调用adb操作类执行截图 保留图片路径 继续分析下一跳:return:'''time.sleep(1.5+random.randint(1,50)/100)self.tempPath = adbOperate.pull_screenshot(self)#是否结束游戏if(self.exit):print('************到达限制次数：%d 自动结束跳跃************' %self.limit_count)returnelse:# 分析下一跳的图片self.show_img()def draw_result(self,src_img,peo_loc,tar_loc,click_tar_loc,click_loc):'''保存识别结果:param src_img:源图像:param peo_loc:人位置:param tar_loc:识别的目标:param click_tar_loc:点击的目标位置:param click_loc:模拟点击的位置:return:'''draw_img=src_img#人draw_img=cv2.circle(draw_img,peo_loc, 3, color=(0, 255, 0), thickness=3)#识别的目标draw_img =cv2.circle(draw_img,tar_loc,3,color=(255,0,0),thickness=3)#点击目标的位置draw_img = cv2.circle(draw_img, click_tar_loc, 3, color=(0, 0, 255), thickness=3)#模拟点击的位置draw_img = cv2.circle(draw_img,click_loc , 6, color=(255, 255, 0), thickness=3)draw_img = cv2.circle(draw_img, click_loc, 12, color=(255, 0, 0), thickness=3)cv2.imwrite(filename=self.tempPath,img=draw_img)

• image_analysis.py

import cv2class IMG_Aanlysis:def __init__(self):#加载小人模板self.template = cv2.imread('.\\template\\peo_template.jpg')#加载中心白点模板self.center_white_template=cv2.imread('.\\template\\white_center_template.png')#加载中心白点模板2self.center_white_template2=cv2.imread('.\\template\\white_center_template1.png')#二值化模板self.template_gray = cv2.cvtColor(self.template, cv2.COLOR_BGR2GRAY)self.template_center_gray = cv2.cvtColor(self.center_white_template, cv2.COLOR_BGR2GRAY)#获取模板的宽高self.w,self. h = self.template_gray.shape[::-1]self.w2, self.h2 = self.template_center_gray.shape[::-1]print('小人模板宽高：'+str(self.w)+':'+str(self.h))print('中心点模板宽高：' + str(self.w2) + ':' + str(self.h2))def get_people_loc(self,img):'''获取小人的位置范围:param img::return:返回小人范围图和小人坐标'''#定义小人位置peo_loc=[0,0]#与模板进行匹配识别 获取识别结果矩形范围min_val, max_val, min_loc, max_loc = self.matchTemplate(img,self.template)#绘制识别结果draw_peo_result = cv2.rectangle(img, max_loc, (max_loc[0] + self.w, max_loc[1] + self.h), color=(255,0,0), lineType=8,thickness=3)#计算小人底部坐标peo_bottom_x=int(max_loc[0]+(self.w)/2)peo_bottom_y=int((max_loc[1]+self.h-19))peo_loc=(peo_bottom_x,peo_bottom_y)draw_peo_result= cv2.circle(draw_peo_result,(peo_bottom_x,peo_bottom_y),2,color=(0,255,0),thickness=3)return (draw_peo_result,peo_loc)def get_target_loc(self,cv_img):'''获取目标位置坐标:param img::return:'''cv_img_gray=cv2.cvtColor(cv_img, cv2.COLOR_BGR2GRAY)min_val2, max_val2, min_loc2, max_loc2 =self.matchTemplate(cv_img_gray,self.template_center_gray)print('模板一匹配中心点 匹配度:%d%% 设定匹配阈值：80%%' % (max_val2 * 100), end=' ')# 计算中心坐标cen_x = int(max_loc2[0] + (self.w2 / 2))cen_y = int(max_loc2[1] + self.h2 / 2)draw_cen_result = cv2.circle(cv_img, (cen_x, cen_y), 2, color=(255, 0, 0), thickness=3)#预防分数干扰 把分数部分设置为全黑色#draw_cen_result[0:500]=[0,0,0]if(max_val2>0.8):print('模板一匹配中心点成功')return (draw_cen_result,(cen_x,cen_y))else:cv_img_gray= cv2.cvtColor(cv_img, cv2.COLOR_BGR2GRAY)center_white_template_gray= cv2.cvtColor(self.center_white_template2, cv2.COLOR_BGR2GRAY)min_val3, max_val3, min_loc3, max_loc3=self.matchTemplate(cv_img_gray,center_white_template_gray)# 计算中心坐标cen_x3 = int(max_loc3[0] + (self.w2 / 2))cen_y3 = int(max_loc3[1] + self.h2 / 2)#print('cen_x3:%d cen_y3:%d' % (cen_x3, cen_y3), end=' ')draw_cen_result = cv2.circle(cv_img, (cen_x3, cen_y3), 2, color=(255, 0, 0), thickness=3)print('模板一匹配中心点失败\n模板二匹配中心点 匹配度:%d%% 设定匹配阈值：80%%' % (max_val3 * 100), end=' ')if (max_val3 > 0.8):print('模板二匹配中心点成功')return (draw_cen_result, (cen_x3, cen_y3))else:print('模板二匹配中心点失败')return (draw_cen_result,(0,0))def matchTemplate(self,img,template):res = cv2.matchTemplate(img, template,cv2.TM_CCOEFF_NORMED)return cv2.minMaxLoc(res)def show_result(self,img):cv2.namedWindow('AnaResult', cv2.WINDOW_GUI_EXPANDED)cv2.imshow('AnaResult', img)cv2.waitKey(0)

import os#定义截图名称
img_name='j.png'
img_out_path='./op_screencap/'def pull_screenshot(self):'''截屏并存放到电脑中:return:'''self.count=  self.count+1self.jump_count=self.jump_count+1if(self.count>30):self.count=1print('将覆盖旧图')os.system('adb shell screencap -p /sdcard/'+img_name)os.system('adb pull /sdcard/j.png '+img_out_path+str(self.count)+'_'+img_name)self.temp_path=img_out_path+str(self.count)+'_'+img_namereturn  self.temp_pathdef jump(distance,click_loc):#由于测试结果发现每次跳跃 系数都会在一定范围波动 因此这里分段设置跳跃系数cons=1.475if(distance<=168):cons=1.57elif(distance>168 and distance<=175):cons=1.5elif(distance>175 and distance<=313):cons=1.62elif(distance>313 and distance<400):cons=1.52elif(distance>=400 and distance<511):cons=1.50elif(distance>=511 and distance<540):cons = 1.49elif(distance>=540 and distance<659):cons=1.45elif(distance>=659 and distance<670):cons=1.47elif(distance>=670 and distance<692):cons=1.45elif(distance>=692 and distance<700):cons=1.38elif(distance>=698):cons=1.40press_time = distance * conspress_time = str(int(press_time))x1=str(click_loc[0])y1=str(click_loc[1])#cmd = 'adb shell input swipe 540 1514 540 1514 ' + str(press_time)cmd = 'adb shell input swipe {0} {1} {2} {3} {4}'cmd= cmd.format(click_loc[0],click_loc[1],click_loc[1],click_loc[0],press_time)#print('[distance:'+str(distance)+'][press_time:'+str(press_time)+']ms [cmd:'+cmd+']')print('跳跃函数：\n-----使用跳跃系数:%1.3f\n-----跳跃距离:%d\n-----按压时间:%s ms' %(cons,int(distance),press_time))os.system(cmd)

import adbOperate
import ImgHelperclass Main:def __init__(self):print('主程序初始化...')imHelper = ImgHelper.ImageHelprt()imHelper.tempPath= adbOperate.pull_screenshot(imHelper)imHelper.show_img()
if __name__ == '__main__':Main()