import cv2
from PIL import Image
name = 'name'
def getface(path):
cap = cv2.VideoCapture(path)
classfier = cv2.CascadeClassifier("/home/coding/workspace/haarcascades/haarcascade_frontalface_alt2.xml")
suc = cap.isOpened() # 是否成功打开
frame_count = 0
out_count = 0
while suc:
frame_count += 1
if out_count > 599: #最多取出多少张
break
suc, frame = cap.read() #读取一帧
params = []
params.append(2) # params.append(1)
grey = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY) #将当前桢图像转换成灰度图像
faceRects = classfier.detectMultiScale(grey, scaleFactor = 1.2, minNeighbors = 3, minSize = (32, 32)) #读取脸部位置
if len(faceRects) > 0: #大于0则检测到人脸
for faceRect in faceRects: #单独框出每一张人脸
x, y, w, h = faceRect
image = frame[y - 10: y + h + 10, x - 10: x + w + 10]
image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) #转为灰度图
img_new = cv2.resize(image,(47,57), interpolation = cv2.INTER_CUBIC) #处理面部的大小
cv2.imwrite('/home/coding/workspace/Data/'+ name +'/%d.jpg' % out_count, img_new, params) #存储到指定目录
out_count += 1
print('成功提取'+ name +'的第%d个脸部'%out_count)
break #每帧只获取一张脸，删除这个即为读出全部面部
cap.release()
cv2.destroyAllWindows()
print('总帧数:', frame_count)
print('提取脸部:',out_count)

if name == 'main':
getface("/home/coding/workspace/Data/Videos/"+ name +".mp4") #参数为视频地址