|
|
|
火焰识别源码，亲测有用，下载后需要搭建opencv QT5的环境。不懂得话可以私聊我
- # -*- coding: cp936 -*-
- import sys
- import threading #线程模块
- import cv2
- import numpy as np
- import time
- import os
- import sys
- from PyQt5 import Qt
- from PyQt5 import QtCore
- from PyQt5.QtGui import QImage, QPixmap,QFont,QIcon
- from PyQt5.QtWidgets import (QApplication,QDialog, QFileDialog, QGridLayout,
- QLabel, QPushButton,QHBoxLayout,QFrame,QWidget,QLineEdit)
-
- font = cv2.FONT_HERSHEY_SIMPLEX #设置字体
-
- class Work(threading.Thread):
- def __init__(self, caller):
- threading.Thread.__init__(self)
- self.caller = caller #父类调用者
- self.isHaveFire = False #全局变量是否检测到火焰
-
- def run(self): #线程启动后自动调用此函数
- cap = cv2.VideoCapture()#初始化VideoCapture类对象
- if(self.caller.video_flag == 1):#标志位为1，对应打开摄像头
- cap = cv2.VideoCapture(0) #打开摄像头
- print("打开摄像头")
- else: #标志位为1，对应打开视频文件
- cap = cv2.VideoCapture(self.caller.video_path)#打开对应路径的视频文件
- print("打开视频文件%s"%self.caller.video_path)
- while(1):
- if(self.caller.flag): #如果视频结束标志位为1，则退出并清空图像
- bgImg = np.ones((640,480,3),np.uint8)*240
- self.showViewImg(self.caller.label, bgImg)
- break
- ret, frame = cap.read() #读取视频或摄像头帧
- if(ret==False):#取帧失败，提示退出循环
- print("摄像头打开失败！" )
- break
- time.sleep(0.001)
- frame = self.fire_detect(frame)#调用火焰检测函数
-
- frameTest = frame.copy()#原图备份
- self.showViewImg(self.caller.label, frame)
-
- cap.release()#释放VideoCapture类对象
- def showViewImg(self,label,img):
- # 提取图像的尺寸和通道, 用于将opencv下的image转换成Qimage
- #self.label.clear()
- channel = 1
- height = width = 1
- try:
- height, width, channel = img.shape
- except:
- channel = 1
- showImg = None
- if channel != 3:
- showImg = cv2.cvtColor(img,cv2.COLOR_GRAY2BGR)
- else:
- showImg = img.copy()
-
- bytesPerLine = 3 * width
- qImg = QImage(showImg.data, width, height, bytesPerLine,
- QImage.Format_RGB888).rgbSwapped()
- # 将Qimage显示出来
- label.setPixmap(QPixmap.fromImage(qImg))
-
- def img_detect(self,img):#加载图片检测火焰
- print("Image Test")
- frame = self.fire_detect(img)#调用检测火焰函数
- self.showViewImg(self.caller.label, frame)
-
-
- def fire_detect(self,frame): #火焰检测函数，核心算法
- self.isHaveFire = False #初始化火焰检测结果标志位False
- redThres = 49 #红色阈值
- sat = 7 #比例系数
- blackImg = np.zeros((frame.shape[0],frame.shape[1],1),np.uint8)#创建原图同大小的黑色图像
- b,g,r = cv2.split(frame)#通道分离
- for i in range(0,frame.shape[0]): #访问所有行
- for j in range(0,frame.shape[1]): #访问所有列
- B = int(b[i,j])#访问第i行，第j列蓝色像素值
- G = int(g[i,j])#访问第i行，第j列绿色像素值
- R = int(r[i,j])#访问第i行，第j列红色像素值
- maxValue = max(max(B,G),R)#求RBG像素最大值
- minValue = min(min(B,G),R)#求RBG像素最小值
- if (R+G+B) == 0:
- break
- S = (1-3.0*minValue/(R+G+B))#计算S值
- if(R>redThres and R>=G and G>=B and S>((255-R)*sat/redThres)):#火焰像素删选
- blackImg[i,j] = 255 #满足火焰像素，黑色图像对应位置变为白色
- else:
- blackImg[i,j] = 0 #不满足火焰像素，黑色图像对应位置仍为黑色
- blackImg = cv2.medianBlur(blackImg,5)#中值滤波滤除小杂讯
- k1=np.ones((5,5), np.uint8)#指定膨胀核大小5*5
- blackImg = cv2.dilate(blackImg, k1, iterations=1)#膨胀
- #查找火焰部分轮廓
- contours,hierarchy = cv2.findContours(blackImg, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
- index = -1
- maxArea = 0
- for i in range (0,len(contours)):#遍历轮廓
- (x0, y0, w0, h0) = cv2.boundingRect(contours[i])#获取轮廓外界矩形
- if(w0>10 and h0>10):#删选外界矩形宽高均大于10
- #cv2.rectangle(frame,(x0,y0),(x0+w0,y0+h0),(0,255,0),2)
- if(w0*h0>maxArea):#比对轮廓面积
- maxArea = w0*h0 #获取最大面积
- index = i #获取最大面积对应的轮廓序号
- if index != -1: #轮廓序号变化了说明没检测到火焰
- area = cv2.contourArea(contours[index])#获取火焰轮廓对应的面积
- cv2.putText(frame,("FireArea=%0.2f"%(area)), (5,20), font, 0.7, (0,255,0), 2)#图片上输出面积
- (x0, y0, w0, h0) = cv2.boundingRect(contours[index])#获取外界矩形
- cv2.rectangle(frame,(x0,y0),(x0+w0,y0+h0),(0,255,0),2)#绘制外界矩形框出火焰区域
- self.isHaveFire = True #检测到火焰标志位为True,对应会播放声音
-
- else: #轮廓序号没变说明没检测到火焰
- cv2.putText(frame,("FireArea=0"), (5,20), font, 0.7, (0,255,0), 2)#火焰面积为0
- return frame #返回最终处理后的图像
-
- if __name__=="__main__":
- pass
|
评分
-
查看全部评分
|
[复制链接]
