OpenCV学习——摄像头人脸识别

151 阅读 0 评论 100 点赞

我是靠谱客的博主虚拟白云，这篇文章主要介绍OpenCV学习——摄像头人脸识别，现在分享给大家，希望可以做个参考。

在学会对图像文件中的人脸进行识别后，还要使用摄像头进行人脸识别，需要先读取摄像头所获取的画面，并将画面进行存储，之后在对存储图像进行人脸识别，并将识别出的人脸存储。

OpenCV学习——图像脸部识别_笨小古的博客-CSDN博客

先学一下脸部储存，需要借助Pillow库

复制代码

import cv2
from PIL import Image

pictPath = r'dataopencvsourcesdatahaarcascadeshaarcascade_frontalface_default.xml'
face_cascade = cv2.CascadeClassifier(pictPath)  # 建立识别文件对象
img = cv2.imread('datag5.jpg')
faces = face_cascade.detectMultiScale(img, scaleFactor=1.1,
                                      minNeighbors=3, minSize=(20, 20))

# 标注右下角底色是黄色
cv2.rectangle(img, (img.shape[1] - 140, img.shape[0] - 20), (img.shape[1], img.shape[0]), (0, 255, 255), -1)

# 标注找到多少人脸
cv2.putText(img, "Finding " + str(len(faces)) + " face", (img.shape[1] - 135, img.shape[0] - 5),
            cv2.FONT_HERSHEY_PLAIN, 1, (255, 0, 0), 1)
# 将人脸框起来，由于有可能找到好几张脸所以用循环绘制出来
num = 1
for (x, y, w, h) in faces:
    cv2.rectangle(img, (x, y), (x + w, y + h), (255, 0, 0), 2)  # 蓝色框住人脸
    filename = "face" + str(num) + ".jpg"  # 建立文件名
    image = Image.open("datag5.jpg")  # PIL模块开启
    imageCrop = image.crop((x, y, x + w, y + h))  # 裁切
    imageResize = imageCrop.resize((150, 150), Image.ANTIALIAS)  # 高质量重制大小 
    imageResize.save(filename)  # 存储文件
    num += 1
cv2.namedWindow("Face", cv2.WINDOW_NORMAL)  # 建立图像对象
cv2.imshow("Face", img)
cv2.waitKey(0)
cv2.destroyAllWindows()

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
import cv2
from PIL import Image

pictPath = r'dataopencvsourcesdatahaarcascadeshaarcascade_frontalface_default.xml'
face_cascade = cv2.CascadeClassifier(pictPath)  # 建立识别文件对象
img = cv2.imread('datag5.jpg')
faces = face_cascade.detectMultiScale(img, scaleFactor=1.1,
                                      minNeighbors=3, minSize=(20, 20))

# 标注右下角底色是黄色
cv2.rectangle(img, (img.shape[1] - 140, img.shape[0] - 20), (img.shape[1], img.shape[0]), (0, 255, 255), -1)

# 标注找到多少人脸
cv2.putText(img, "Finding " + str(len(faces)) + " face", (img.shape[1] - 135, img.shape[0] - 5),
            cv2.FONT_HERSHEY_PLAIN, 1, (255, 0, 0), 1)
# 将人脸框起来，由于有可能找到好几张脸所以用循环绘制出来
num = 1
for (x, y, w, h) in faces:
    cv2.rectangle(img, (x, y), (x + w, y + h), (255, 0, 0), 2)  # 蓝色框住人脸
    filename = "face" + str(num) + ".jpg"  # 建立文件名
    image = Image.open("datag5.jpg")  # PIL模块开启
    imageCrop = image.crop((x, y, x + w, y + h))  # 裁切
    imageResize = imageCrop.resize((150, 150), Image.ANTIALIAS)  # 高质量重制大小 
    imageResize.save(filename)  # 存储文件
    num += 1
cv2.namedWindow("Face", cv2.WINDOW_NORMAL)  # 建立图像对象
cv2.imshow("Face", img)
cv2.waitKey(0)
cv2.destroyAllWindows()

（左边第二位是中国台湾著名计算机专家，IT图书知名作者洪锦魁，曾出版了多部Python学习书籍）

下面使用笔记本电脑摄像头进行人脸识别

复制代码

1
cap = VideoCapture(n)  # 笔记本电脑上内置摄像头，n是0

复制代码

import cv2
from PIL import Image

pictPath = r'dataopencvsourcesdatahaarcascadeshaarcascade_frontalface_default.xml'
face_cascade = cv2.CascadeClassifier(pictPath)  # 建立识别文件对象
cv2.namedWindow("Photo")
cap = cv2.VideoCapture(0)  # 开启摄像头
while (cap.isOpened()):  # 摄像头开启后执行
    ret, img = cap.read()  # 读取图像;ret是布尔值，摄像头拍摄成功便返回True,img是摄像头所拍的图像对象
    cv2.imshow("Photo", img)  # 显示图像在OpenCV窗口
    if ret == True:  # 如果图像读取成功
        key = cv2.waitKey(200)  # 每0.2秒检查一次
        if key == ord("a") or key == ord("A"):
            cv2.imwrite("photo.jpg", img)  # 将图像写入photo.jpg
            break
cap.release()  # 关闭摄像头

faces = face_cascade.detectMultiScale(img, scaleFactor=1.1,
                                      minNeighbors=3, minSize=(20, 20))

# 标注右下角底色是黄色
cv2.rectangle(img, (img.shape[1] - 140, img.shape[0] - 20), (img.shape[1], img.shape[0]), (0, 255, 255), -1)

# 标注找到多少人脸
cv2.putText(img, "Finding " + str(len(faces)) + " face", (img.shape[1] - 135, img.shape[0] - 5),
            cv2.FONT_HERSHEY_PLAIN, 1, (255, 0, 0), 1)
# 将人脸框起来，由于有可能找到好几张脸所以用循环绘制出来
for (x, y, w, h) in faces:
    cv2.rectangle(img, (x, y), (x + w, y + h), (255, 0, 0), 2)  # 蓝色框住人脸
    image = Image.open("photo.jpg")  # PIL模块开启
    imageCrop = image.crop((x, y, x + w, y + h))  # 裁切
    imageResize = imageCrop.resize((150, 150), Image.ANTIALIAS)  # 高质量重制大小
    imageResize.save("faceout.jpg")  # 存储文件

cv2.namedWindow("FaceRecognition", cv2.WINDOW_NORMAL)  # 建立图像对象
cv2.imshow("FaceRecognition", img)
cv2.waitKey(0)
cv2.destroyAllWindows()

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
import cv2
from PIL import Image

pictPath = r'dataopencvsourcesdatahaarcascadeshaarcascade_frontalface_default.xml'
face_cascade = cv2.CascadeClassifier(pictPath)  # 建立识别文件对象
cv2.namedWindow("Photo")
cap = cv2.VideoCapture(0)  # 开启摄像头
while (cap.isOpened()):  # 摄像头开启后执行
    ret, img = cap.read()  # 读取图像;ret是布尔值，摄像头拍摄成功便返回True,img是摄像头所拍的图像对象
    cv2.imshow("Photo", img)  # 显示图像在OpenCV窗口
    if ret == True:  # 如果图像读取成功
        key = cv2.waitKey(200)  # 每0.2秒检查一次
        if key == ord("a") or key == ord("A"):
            cv2.imwrite("photo.jpg", img)  # 将图像写入photo.jpg
            break
cap.release()  # 关闭摄像头

faces = face_cascade.detectMultiScale(img, scaleFactor=1.1,
                                      minNeighbors=3, minSize=(20, 20))

# 标注右下角底色是黄色
cv2.rectangle(img, (img.shape[1] - 140, img.shape[0] - 20), (img.shape[1], img.shape[0]), (0, 255, 255), -1)

# 标注找到多少人脸
cv2.putText(img, "Finding " + str(len(faces)) + " face", (img.shape[1] - 135, img.shape[0] - 5),
            cv2.FONT_HERSHEY_PLAIN, 1, (255, 0, 0), 1)
# 将人脸框起来，由于有可能找到好几张脸所以用循环绘制出来
for (x, y, w, h) in faces:
    cv2.rectangle(img, (x, y), (x + w, y + h), (255, 0, 0), 2)  # 蓝色框住人脸
    image = Image.open("photo.jpg")  # PIL模块开启
    imageCrop = image.crop((x, y, x + w, y + h))  # 裁切
    imageResize = imageCrop.resize((150, 150), Image.ANTIALIAS)  # 高质量重制大小
    imageResize.save("faceout.jpg")  # 存储文件

cv2.namedWindow("FaceRecognition", cv2.WINDOW_NORMAL)  # 建立图像对象
cv2.imshow("FaceRecognition", img)
cv2.waitKey(0)
cv2.destroyAllWindows()