表情识别（一）——使用Dlib、opencv和Python识别面部特征

最新推荐文章于 2024-06-21 01:46:42 发布

客院载论

最新推荐文章于 2024-06-21 01:46:42 发布

阅读量4.4k

点赞数 5

分类专栏：表情识别文章标签： dlib opencv 人脸识别

本文链接：https://blog.csdn.net/Blackoutdragon/article/details/119764654

版权

表情识别专栏收录该内容

3 篇文章 6 订阅

订阅专栏

文章目录

- - 基本步骤

基本步骤

1、定位图片中的脸

面部检测可以使用很多方式实现，比如说OpenCV内嵌的Haar级联，预训练好的HOG+ 先行SVM对象检测，或者使用深度学习算法进行面部识别。无论什么方法，我们最终都是要获得标定面部的选区

2、在面部ROI中检测出关键的面部结构

主要给出对应面部选区，我们就可以进行面部特征点的检测。有很多不同的面部特征点检测器，但是所有的方法基本上都是针对以下几个器官：嘴、左右眼睫毛、左右眼、鼻子、下巴等。
Dlib库中常用的面部特征检测器是One millisecond face alignment with an ensemble of regression trees。这个方法使用了一个人工标注的测试集，标注的内容是围绕面部特征的特定的坐标，坐标表示的是像素点之间的距离。有了这个训练集，就可以训练出来一个集成的回归树，用来检测面部的特征。这个特征检测器的可以进行实时高精度检测。
如果想要的更加深入的了解这个技术，可以通过连接，读相关的文章，配合Dlib的官方文档。
文章的连接
Dlib官方文档
在Dlib库中的预先训练好的面部特征检测是针对人脸结构上的68个特征点，分布如下。

在这里插入图片描述

什么是ROI

图像处理中，从被处理图像以方框、圆、椭圆等不规则多边形方式勾勒出的需要处理的区域，成为感兴趣区域，ROI。

补充函数rect_to_bb，将rect转成坐标点

描述：将检测器检测出来的rect转换成具体的长方形框的坐标点
原理：detecor返回的值是rect，数据的形式是（x,y,height,width）

def rect_to_bb(rect):
	# take a bounding predicted by dlib and convert it
	# to the format (x, y, w, h) as we would normally do
	# with OpenCV
	x = rect.left()
	y = rect.top()
	w = rect.right() - x
	h = rect.bottom() - y
	# return a tuple of (x, y, w, h)
	return (x, y, w, h)

补充函数shapetonp

描述：将包含的68个面部区域的坐标点的shape转为numpy数组

def shape_to_np(shape, dtype="int"):
	# initialize the list of (x, y)-coordinates
	coords = np.zeros((68, 2), dtype=dtype)
	# loop over the 68 facial landmarks and convert them
	# to a 2-tuple of (x, y)-coordinates
	for i in range(0, 68):
		coords[i] = (shape.part(i).x, shape.part(i).y)
	# return the list of (x, y)-coordinates
	return coords

补充函数resive

描述：将图片按照要求设定大小
参数：image是cv2.imread的对象

  	 width和height是新指定的大小参数

def resize(image, width=None, height=None, inter=cv2.INTER_AREA):
    # initialize the dimensions of the image to be resized and
    # grab the image size
    dim = None
    (h, w) = image.shape[:2]

    # if both the width and height are None, then return the
    # original image
    if width is None and height is None:
        return image

    # check to see if the width is None
    if width is None:
        # calculate the ratio of the height and construct the
        # dimensions
        r = height / float(h)
        dim = (int(w * r), height)

    # otherwise, the height is None
    else:
        # calculate the ratio of the width and construct the
        # dimensions
        r = width / float(w)
        dim = (width, int(h * r))

    # resize the image
    resized = cv2.resize(image, dim, interpolation=inter)

    # return the resized image
    return resized

主要代码

导入相关的包

# import the necessary packages
import numpy as np
import argparse
import dlib
import cv2

初始化面部检测器和面部特征预测器

# initialize dlib's face detector (HOG-based) and then create
# the facial landmark predictor
# 初始化基于HOG预测的预先训练好的检测器
detector = dlib.get_frontal_face_detector()
# 使用的shape-predictor去加载的下载的面部特征训练器
# 括号里的是检测器的路径
predictor = dlib.shape_predictor("the path of the detector")

打开图片并读取，将之转换为的灰度图片，固定大小

# 使用opencv打开图片
image = cv2.imread("1.jpg")
# 统一图片的大小
image = imutils.resize(image,width = 500)
# 将图片转为灰度图片，将BGR图片转为灰度图片
gray = cv2.cvtColor(image,cv2.COLOR_BGR2GRAY)

具体resize函数

def resize(image, width=None, height=None, inter=cv2.INTER_AREA):
    # initialize the dimensions of the image to be resized and
    # grab the image size
    dim = None
    (h, w) = image.shape[:2]

    # if both the width and height are None, then return the
    # original image
    if width is None and height is None:
        return image

    # check to see if the width is None
    if width is None:
        # calculate the ratio of the height and construct the
        # dimensions
        r = height / float(h)
        dim = (int(w * r), height)

    # otherwise, the height is None
    else:
        # calculate the ratio of the width and construct the
        # dimensions
        r = width / float(w)
        dim = (width, int(h * r))

    # resize the image
    resized = cv2.resize(image, dim, interpolation=inter)

    # return the resized image
    return resized

imread的具体的输出，每一个像素点是以（r，g，b）的形式进行保存的，结果如下

在这里插入图片描述
原图片的shape输出，对应的是heightweightchannel，总共是rgb三个颜色的通道，像素点是711*474

修改之后的图片尺寸

在这里插入图片描述

转换之后的灰度图片，仅仅只有一个单通道

在这里插入图片描述

调用加载好的检测器，对目标进行检测

第一个参数是需要检测的图片
第二个参数是图片的层数，这里是单层图片，只有一个灰度层

# detect face in the grayscale face
# detecting the bounding box of faces in our image
# the second parameter is the number of image pyramid layer
# prior applying the detector we must upscaling the image
rects = detector(gray,1)

在这里插入图片描述

rects的结果是的坐标和weight和height两对参数

遍历所有是别人出来的人脸

for (i,rect) in enumerate(rects):
    # i对应的是目标的索引，rect对应是每一个框的起始点坐标和长宽
    # 定位人脸的关键点，返回的定位之后的68个关键点的位置
    shape = predictor(gray,rect)
    # shape是输出之后坐标点，是（68，2），68个点，每个点都是二维的，将所有的坐标点转为numpy数组
    shape = face_utils.shape_to_np(shape)

    # 将rect检测人脸的边框转为绘制矩形框的具体位置
    (x,y,w,h) = face_utils.rect_to_bb(rect)
    # 绘制人脸的矩形框
    cv2.rectangle(image,(x,y),(x+w,y+h),(0,255,0),2)

    # 设置矩形框的文字部分
    cv2.putText(image,"Face #{}".format(i+1),(x-10,y-10),
                cv2.FONT_HERSHEY_SIMPLEX,0.5,(0,255,0),2)

    # 循环遍历所有的点，将之在原图上进行标注
    for (x,y) in shape:
        cv2.circle(image,(x,y),1,(0,0,255),-1)

具体的函数face_utils.shape_to_np的具体代码，将shape的输出结果转为numpy数组

def shape_to_np(shape, dtype="int"):
	# initialize the list of (x, y)-coordinates
	coords = np.zeros((68, 2), dtype=dtype)
	# loop over the 68 facial landmarks and convert them
	# to a 2-tuple of (x, y)-coordinates
	for i in range(0, 68):
		coords[i] = (shape.part(i).x, shape.part(i).y)
	# return the list of (x, y)-coordinates
	return coords

输出修改之后的图片

# show the output image with the face detections + facial landmarks
cv2.imshow("Output",image)
cv2.waitKey(0)

最终的代码

# import the necessary packages

# import argparse
import cv2
import dlib

import imutils
# the package below from the writer
from imutils import face_utils


# intialize dlib face detector adn then create the facial landmark predictor
detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")

# load the image,resize it and convert it into grayscale
# resice the size of the image into 500 width
# image = cv2.imread(args["image"])
image = cv2.imread("1.jpg")
image = imutils.resize(image,width = 500)
gray = cv2.cvtColor(image,cv2.COLOR_BGR2GRAY)

# detect face in the grayscale face
# detecting the bounding box of faces in our image
# the second parameter is the number of image pyramid layer
# prior applying the detector we must upscaling the image
rects = detector(gray,1)


# Given the coordinates of the face in the image
# loop over the face detections
for (i,rect) in enumerate(rects):
    # determine the facial landmarks for the face region
    # convert the coordiantes of the facial landmark to numpy array
    # predictor is to detect the facila landmark
    shape = predictor(gray,rect)
    # convert the dlib objects to a numpy array
    shape = face_utils.shape_to_np(shape)

    # convert dlib's rectangle to a OpenCV-style bounding box(x,y,w,h)
    # then draw the face bounding box
    (x,y,w,h) = face_utils.rect_to_bb(rect)
    cv2.rectangle(image,(x,y),(x+w,y+h),(0,255,0),2)

    # show the face number
    cv2.putText(image,"Face #{}".format(i+1),(x-10,y-10),
                cv2.FONT_HERSHEY_SIMPLEX,0.5,(0,255,0),2)

    # loop over the (x,y)-coordinates for the facial lanmarks
    # and draw the on the image
    for (x,y) in shape:
        cv2.circle(image,(x,y),1,(0,0,255),-1)

# show the output image with the face detections + facial landmarks
cv2.imshow("Output",image)
cv2.waitKey(0)