计算机视觉人脸识别 Face Recognition

人脸识别并动态贴图 实现抖音动态小表情

import numpy as np
import cv2
import time
import datetime
from PIL import Image

cap = cv2.VideoCapture(0)

''' 人脸识别 '''
def getface(img):
    # 人脸识别数据
    face_cascade = cv2.CascadeClassifier(
        'C:/ProgramData/Anaconda3/envs/pra/Lib/site-packages/cv2/data/haarcascade_frontalface_default.xml')
    # 人眼识别数据
    eye_cascade = cv2.CascadeClassifier(
        'C:/ProgramData/Anaconda3/envs/pra/Lib/site-packages/cv2/data/haarcascade_eye.xml')
    # 二值化,变为灰度图
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    # 获取人脸识别数据
    faces = face_cascade.detectMultiScale(gray, 1.3, 5)
    for (x, y, w, h) in faces:
        # 绘画人脸识别数据
        img = cv2.rectangle(img, (x, y), (x + w, y + h), (255, 0, 0), 2)
        # 根据人脸识别数据添加头像
        img = christmas(img, x, y, w, h)
    return img


def christmas(img, x, y, w, h):
    im = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
    # 你的贴纸地址
    mark = Image.open("hat/111.png")
    height = int(w * 987 / 1024)
    mark = mark.resize((w, height))
    layer = Image.new('RGBA', im.size, (0, 0, 0, 0))
    layer.paste(mark, (x, y - height + 100))
    out = Image.composite(layer, im, layer)
    img = cv2.cvtColor(np.asarray(out), cv2.COLOR_RGB2BGR)
    return img


videoWriter = cv2.VideoWriter('testwrite.avi', cv2.VideoWriter_fourcc(*'MJPG'), 15, (1000, 563))

while (cap.isOpened()):
    ret, frame = cap.read()
    if ret == True:
        # 从新定义图片大小
        img = cv2.resize(frame, (1000, 563))

        # 添加录像时间
        # img = addtime(img)
        # 实时识别
        img = getface(img)
        # 视频显示
        cv2.imshow('frame', img)
        # 保存视频
        videoWriter.write(img)
        if cv2.waitKey(10) & 0xFF == ord('q'):
            print("退出视频")
            break
    else:
        break

cap.release()
videoWriter.release()
cv2.destroyAllWindows()