1.python實時調取本地攝像頭
import numpy as np import cv2 cap = cv2.VideoCapture(0) #參數為0時調用本地攝像頭；url連接調取網絡攝像頭；文件地址獲取本地視頻 while(True): ret,frame=cap.read() #灰度化 gray=cv2.cvtColor(frame,cv2.COLOR_BGR2GRAY) cv2.imshow('frame',gray) #普通圖片 cv2.imshow('frame',frame) if cv2.waitKey(1)&0xFF==ord('q'): break cap.release() cv2.destroyAllWindows()
2.opencv代碼
# -*- coding: utf-8 -*- """ Spyder Editor This is a temporary script file. """ #設置工作路徑 import os os.chdir('E:\\0yfl\\SH-spyder-workspace\\') os.path.abspath('.') import numpy as np import cv2 #1.1讀取圖片imread；展示圖片imshow；導出圖片imwrite #只是灰度圖片 img=cv2.imread('Myhero.jpg',cv2.IMREAD_GRAYSCALE) #彩色圖片 img=cv2.imread('Myhero.jpg',cv2.IMREAD_COLOR) #彩色以及帶有透明度 img=cv2.imread('Myhero.jpg',cv2.IMREAD_UNCHANGED) print(img) #設置窗口可自動調節大小 cv2.namedWindow('image',cv2.WINDOW_NORMAL) cv2.imshow('image',img) k=cv2.waitKey(0) #如果輸入esc if k==27: #exit cv2.destroyAllWindows #如果輸入s elif k==ord('s'): #save picture and exit cv2.imwrite('Myhero_out.png',img) cv2.destroyAllWindows() #1.2視頻讀取 #打開內置攝像頭 cap=cv2.VideoCapture(0) #打開視頻 cap=cv2.VideoCapture('why.mp4') #或者視頻每秒多少幀的數據 fps=cap.get(5) i=0 while(True): #讀取一幀 ret,frame=cap.read() #轉化為灰圖 gray=cv2.cvtColor(frame,cv2.COLOR_BGR2GRAY) #設置導出文件名編號 i = i + 1 #每1s導出一張 if i/fps==int(i/fps): #導出文件名為why+編號+.png #若想要導出灰圖，則將下面frame改為gray即可 cv2.imwrite("why"+str(int(i/fps))+".png",frame) #讀完之后結束退出 if cv2.waitKey(1)==ord('q'): break cap.release() cv2.destoryAllWindows() #1.3圖像像素修改 rangexmin=100 rangexmax=120 rangeymin=90 rangeymax=100 img=cv2.imread('Myhero.jpg',0) img[rangexmin:rangexmax,rangeymin:rangeymax]=[[255]*(rangeymax-rangeymin)]*(rangexmax-rangexmin) cv2.imwrite('Myhero_out2.png',img) #拆分以及合並圖像通道1 b,g,r=cv2.split(img) img=cv2.merge(b,g,r) #png轉eps，不過非常模糊 from matplotlib import pyplot as plt img=cv2.imread('wechat1.png',cv2.IMREAD_COLOR) plt.imsave('wechat_out.eps',img) #圖像按比例混合 img1=cv2.imread('Myhero.jpg',cv2.IMREAD_COLOR) img2=cv2.imread('Myhero_out.png',cv2.IMREAD_COLOR) dst=cv2.addWeighted(img1,0.5,img2,0.5,0) cv2.imwrite("Myhero_combi.jpg",dst) #1.4按位運算 #加載圖像 img1=cv2.imread("Myhero.jpg") img2=cv2.imread("why1.png") #后面那張圖更大 rows,cols,channels=img1.shape ROI=img2[0:rows,0:cols] #做一個ROI為圖像的大小 img2gray=cv2.cvtColor(img1,cv2.COLOR_BGR2GRAY) #小於175的改為0，大於175的賦值為255 ret,mask=cv2.threshold(img2gray,175,255,cv2.THRESH_BINARY) cv2.imwrite("Myhero_mask.jpg",mask) #255-mask=mask_inv mask_inv=cv2.bitwise_not(mask) cv2.imwrite("Myhero_mask_inv.jpg",mask_inv) #在mask白色區域顯示成ROI，背景圖片 img2_bg=cv2.bitwise_and(ROI,ROI,mask=mask) cv2.imwrite("Myhero_pic2_backgroud.jpg",img2_bg) #除了mask以外的區域都顯示成img1，前景圖片 img1_fg=cv2.bitwise_and(img1,img1,mask=mask_inv) cv2.imwrite("Myhero_pic2_frontgroud.jpg",img1_fg) #前景圖片加上背景圖片 dst = cv2.add(img2_bg,img1_fg) img2[0:rows, 0:cols ] = dst cv2.imwrite("Myhero_pic2_addgroud.jpg",dst) #finished #構建淹膜方法2 #截取幀 ret,frame=cap.read() #轉換到HSV hsv=cv2.cvtColor(frame,cv2.COLOR_BGR2HSV) #設定藍色的閾值 lower_blue=np.array([110,50,50]) upper_blue=np.array([130,255,255]) #根據閾值構建掩模 mask=cv2.inRange(hsv,lower_blue,upper_blue) #對原圖像和掩模進行位運算 res=cv2.bitwise_and(frame,frame,mask=mask) #圖片放縮，用的插值方法，所以不會損害清晰度 res=cv2.resize(img1,None,fx=2,fy=2,interpolation=cv2.INTER_CUBIC) cv2.imwrite("Myhero_bigger.jpg",res) #第二種插值方法 height,width=img.shape[:2] res=cv2.resize(img,(2*width,2*height),interpolation=cv2.INTER_CUBIC) #edge現實圖片中不好用，人工畫的圖片還可以 img = cv2.imread('why3.png',0) edges = cv2.Canny(img,50,100) cv2.imwrite("why3_edge.png",edges) #識別輪廓，並保存輪廓點contours img=cv2.imread('why129.png') imgray=cv2.imread('why129.png',cv2.IMREAD_GRAYSCALE) ret,thresh = cv2.threshold(imgray,127,255,0) cv2.imwrite("2.jpg",thresh) image, contours, hierarchy = cv2.findContours(thresh,cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE) img = cv2.drawContours(img, contours, -1, (0,255,0), 3) cv2.imwrite("3.jpg",img) #輪廓 img = cv2.imread('why3.png',0) ret,thresh = cv2.threshold(img,127,255,0) contours,hierarchy = cv2.findContours(thresh, 1, 2) cnt = contours[0] #近似輪廓 epsilon = 0.1*cv2.arcLength(cnt,True) approx = cv2.approxPolyDP(cnt,epsilon,True) img = cv2.drawContours(img, approx, -1, (0,255,0), 3) cv2.imwrite("4.jpg",img) from matplotlib import pyplot as plt #圖像識別/匹配 img_rgb = cv2.imread('why174.png') img_gray = cv2.cvtColor(img_rgb, cv2.COLOR_BGR2GRAY) img2=img_gray.copy() template = cv2.imread('0temp.png',0) w, h = template.shape[::-1] #共有六種識別方法 methods = ['cv2.TM_CCOEFF', 'cv2.TM_CCOEFF_NORMED', 'cv2.TM_CCORR', 'cv2.TM_CCORR_NORMED', 'cv2.TM_SQDIFF', 'cv2.TM_SQDIFF_NORMED'] for meth in methods: img = img2.copy() #eval返回某個式子的計算結果 method = eval(meth) #下面使用匹配方法 res = cv2.matchTemplate(img,template,method) 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 bottom_right = (top_left[0] + w, top_left[1] + h) #畫矩形把他框出來 cv2.rectangle(img,top_left, bottom_right, 255, 2) plt.subplot(121),plt.imshow(res,cmap = 'gray') plt.title('Matching Result'), plt.xticks([]), plt.yticks([]) plt.subplot(122),plt.imshow(img,cmap = 'gray') plt.title('Detected Point'), plt.xticks([]), plt.yticks([]) plt.suptitle(meth) plt.show() #這個匹配結果太差 #選取3，5，6的匹配方式會稍微好點：cv2.TM_CCORR；cv2.TM_SQDIFF，cv2.TM_SQDIFF_NORMED #視頻人臉識別 #https://blog.csdn.net/wsywb111/article/details/79152425 import cv2 from PIL import Image cap=cv2.VideoCapture("why.mp4") #告訴Opencv使用人臉識別分類器 classfier=cv2.CascadeClassifier("E:\\0yfl\\opencv-master\\data\\haarcascades\\haarcascade_frontalface_alt2.xml") count=0 while cap.isOpened(): ret,frame=cap.read() if not ret: break grey=cv2.cvtColor(frame,cv2.COLOR_BGR2GRAY) faceRect=classfier.detectMultiScale(grey,scaleFactor=1.2, minNeighbors=3, minSize=(32, 32)) if len(faceRect)>0: count=count+1 print(count) #137這種程度可以識別，111沒有成功識別，大概是側臉的緣故 #截出人臉 image_name="why111.png" frame=cv2.imread(image_name,0) if not (frame is None): #導入測試集 classfier=cv2.CascadeClassifier("E:\\0yfl\\opencv-master\\data\\haarcascades\\haarcascade_frontalface_alt2.xml") #使用測試集導出人臉的位置，存在faceRect中，可以檢測多張人臉 faceRect=classfier.detectMultiScale(frame,scaleFactor=3.0, minNeighbors=3, minSize=(32, 32)) count=0 for (x1,y1,w,h) in faceRect: count=count+1 #截取上述圖片的人臉部分並保存每一張識別出的人臉 Image.open(image_name).crop((x1,y1,x1+w,y1+h)).save(image_name.split(".")[0]+"_face_"+str(count)+".png") if count==0: print ("No face detected!") else: print ("Picture "+ image_name +" is not exist in "+os.path.abspath(".")) #人臉上畫出矩形 from PIL import Image,ImageDraw image_name="why111.png" frame=cv2.imread(image_name,0) if not (frame is None): classfier=cv2.CascadeClassifier("E:\\0yfl\\opencv-master\\data\\haarcascades\\haarcascade_frontalface_alt2.xml") faceRect=classfier.detectMultiScale(frame,scaleFactor=3.0, minNeighbors=3, minSize=(32, 32)) #畫框框 img = Image.open(image_name) draw_instance = ImageDraw.Draw(img) count=0 for (x1,y1,w,h) in faceRect: draw_instance.rectangle((x1,y1,x1+w,y1+h), outline=(255, 0,0)) img.save('drawfaces_'+image_name) count=count+1 if count==0: print ("No face detected!") else: print ("Picture "+ image_name +" is not exist in "+os.path.abspath(".")) #detectFaces()返回圖像中所有人臉的矩形坐標（矩形左上、右下頂點） #使用haar特征的級聯分類器haarcascade_frontalface_default.xml，在haarcascades目錄下還有其他的訓練好的xml文件可供選擇。 #注：haarcascades目錄下訓練好的分類器必須以灰度圖作為輸入。 from PIL import Image,ImageDraw image_name="why63.png" frame=cv2.imread(image_name,0) if not (frame is None): classfier=cv2.CascadeClassifier("E:\\0yfl\\opencv-master\\data\\haarcascades\\haarcascade_fullbody.xml") faceRect=classfier.detectMultiScale(frame,scaleFactor=3.0, minNeighbors=3, minSize=(32, 32)) #畫框框 img = Image.open(image_name) draw_instance = ImageDraw.Draw(img) count=0 for (x1,y1,w,h) in faceRect: draw_instance.rectangle((x1,y1,x1+w,y1+h), outline=(255, 0,0)) img.save('drawfaces_'+image_name) count=count+1 if count==0: print ("No face detected!") else: print ("Picture "+ image_name +" is not exist in "+os.path.abspath("."))