# -*-coding:utf8-*-#

import os
import cv2
from PIL import Image, ImageDraw
from datetime import datetime

"""
 detectFaces()返回图像中所有人脸的矩形坐标(矩形左上、右下顶点)
 使用haar特征的级联分类器haarcascade_frontalface_default.xml,在haarcascades目录下还有其他的训练好的xml文件可供选择。
 注:haarcascades目录下训练好的分类器必须以灰度图作为输入。
 分类器 https://github.com/opencv/opencv/tree/master/data/haarcascades
 安装模块:pip install Pillow pip install opencv-python
"""


def detectFaces(image_name):
 img = cv2.imread(image_name)
 face_cascade = cv2.CascadeClassifier(os.getcwd()+"\\haarcascade\\haarcascade_frontalface_alt.xml")
 if img.ndim == 3:
 gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
 else:
 gray = img # if语句:如果img维度为3,说明不是灰度图,先转化为灰度图gray,如果不为3,也就是2,原图就是灰度图

 faces = face_cascade.detectMultiScale(gray, 1.2, 5) # 1.3和5是特征的最小、最大检测窗口,它改变检测结果也会改变
 result = []
 for (x, y, width, height) in faces:
 result.append((x, y, x + width, y + height))
 return result


# 保存人脸图
def saveFaces(image_name):
 faces = detectFaces(image_name)
 if faces:
 # 将人脸保存在save_dir目录下。
 # Image模块:Image.open获取图像句柄,crop剪切图像(剪切的区域就是detectFaces返回的坐标),save保存。
 save_dir = image_name.split('.')[0] + "_faces"
 os.mkdir(save_dir)
 count = 0
 for (x1, y1, x2, y2) in faces:
 file_name = os.path.join(save_dir, str(count) + ".jpg")
 Image.open(image_name).crop((x1, y1, x2, y2)).save(file_name)
 count += 1


# 在原图像上画矩形,框出所有人脸。
# 调用Image模块的draw方法,Image.open获取图像句柄,ImageDraw.Draw获取该图像的draw实例,然后调用该draw实例的rectangle方法画矩形(矩形的坐标即
# detectFaces返回的坐标),outline是矩形线条颜色(B,G,R)。
# 注:原始图像如果是灰度图,则去掉outline,因为灰度图没有RGB可言。drawEyes、detectSmiles也一样。
def drawFaces(path, image_name):
 faces = detectFaces(path+image_name)
 if faces:
 img = Image.open(path+image_name)
 draw_instance = ImageDraw.Draw(img)
 for (x1, y1, x2, y2) in faces:
 draw_instance.rectangle((x1, y1, x2, y2), outline=(255, 0, 0))
 img.save(path+'drawfaces_' + image_name)


# 检测眼睛,返回坐标
# 由于眼睛在人脸上,我们往往是先检测出人脸,再细入地检测眼睛。故detectEyes可在detectFaces基础上来进行,代码中需要注意"相对坐标"。
# 当然也可以在整张图片上直接使用分类器,这种方法代码跟detectFaces一样,这里不多说。
def detectEyes(image_name):
 eye_cascade = cv2.CascadeClassifier(os.getcwd()+"\\haarcascade\\haarcascade_eye.xml")
 faces = detectFaces(image_name)

 img = cv2.imread(image_name)
 gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
 result = []
 for (x1, y1, x2, y2) in faces:
 roi_gray = gray[y1:y2, x1:x2]
 eyes = eye_cascade.detectMultiScale(roi_gray, 1.3, 2)
 for (ex, ey, ew, eh) in eyes:
 result.append((x1 + ex, y1 + ey, x1 + ex + ew, y1 + ey + eh))
 return result


# 在原图像上框出眼睛.
def drawEyes(image_name):
 eyes = detectEyes(image_name)
 if eyes:
 img = Image.open(image_name)
 draw_instance = ImageDraw.Draw(img)
 for (x1, y1, x2, y2) in eyes:
 draw_instance.rectangle((x1, y1, x2, y2), outline=(0, 0, 255))
 img.save('draweyes_' + image_name)


# 检测笑脸
def detectSmiles(image_name):
 img = cv2.imread(image_name)
 smiles_cascade = cv2.CascadeClassifier(os.getcwd()+"\\haarcascade\\haarcascade_smile.xml")
 if img.ndim == 3:
 gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
 else:
 gray = img # if语句:如果img维度为3,说明不是灰度图,先转化为灰度图gray,如果不为3,也就是2,原图就是灰度图

 smiles = smiles_cascade.detectMultiScale(gray, 4, 5)
 result = []
 for (x, y, width, height) in smiles:
 result.append((x, y, x + width, y + height))
 return result


# 在原图像上框出笑脸
def drawSmiles(image_name):
 smiles = detectSmiles(image_name)
 if smiles:
 img = Image.open(image_name)
 draw_instance = ImageDraw.Draw(img)
 for (x1, y1, x2, y2) in smiles:
 draw_instance.rectangle((x1, y1, x2, y2), outline=(100, 100, 0))
 img.save('drawsmiles_' + image_name)


if __name__ == '__main__':
 time1 = datetime.now()
 result = detectFaces(os.getcwd()+"\\images\\heying.jpg")
 time2 = datetime.now()
 print("耗时:" + str(time2 - time1))
 if len(result) > 0:
 print("有人存在!!---》人数为:" + str(len(result)))
 else:
 print('视频图像中无人!!')

 drawFaces(os.getcwd()+"\\images\\", "heying.jpg")
 # saveFaces(os.getcwd()+"\\images\\heying.jpg")
 # drawSmiles('img/people.jpg') # 极其不准确
 # drawEyes('people.jpg') # 有问题报错


"""
上面的代码将眼睛、人脸、笑脸在不同的图像上框出,如果需要在同一张图像上框出,改一下代码就可以了。
总之,利用opencv里训练好的haar特征的xml文件,在图片上检测出人脸的坐标,利用这个坐标,我们可以将人脸区域剪切保存,也可以在原图上将人脸框出。剪切保存人脸以及用矩形工具框出人脸,本程序使用的是PIL里的Image、ImageDraw模块。
此外,opencv里面也有画矩形的模块,同样可以用来框出人脸。
"""