OpenCV—快速人脸检测（Haar分类器）

1、OpenCV人脸检测的方法

在OpenCV中主要使用了两种特征（即两种方法）进行人脸检测，Haar特征和LBP特征。在OpenCV中，使用已经训练好的XML格式的分类器进行人脸检测。

下载地址：https://github.com/opencv/opencv

在OpenCV的安装目录下的sources文件夹里的data文件夹里可以看到下图所示的内容：

上图中文件夹的名字“haarcascades”、“hogcascades”和“lbpcascades”分别表示通过“haar”、“hog”和“lbp”三种不同的特征而训练出的分类器：即各文件夹里的文件。"haar"特征主要用于人脸检测，“hog”特征主要用于行人检测，“lbp”特征主要用于人脸识别。打开“haarcascades”文件夹，如下图所示

图中的XML文件即是我们人脸检测所需要的分类器文件。在实际使用中，推荐使用上图中被标记的“haarcascade_frontalface_alt2.xml”分类器文件，准确率和速度都比较好。 

OpenCV支持的目标检测的方法是利用样本的Haar特征进行的分类器训练，得到的级联boosted分类器（Cascade Classification）。OpenCV2之后的C++接口除了Haar特征以外也可以使用LBP特征。

2、OpenCV中的人脸检测的类

在OpenCV中，使用类“CascadeClassifier”进行人脸检测

# encoding: utf-8
# module cv2.cv2
# from D:\JetBrains\PyCharm 2019.1.3\PycharmProjects\Study\venv\lib\site-packages\cv2\cv2.cp36-win_amd64.pyd
# by generator 1.147
""" Python wrapper for OpenCV. """

# imports
import cv2.cv2 as  # D:\JetBrains\PyCharm 2019.1.3\PycharmProjects\Study\venv\lib\site-packages\cv2\cv2.cp36-win_amd64.pyd
import cv2.Error as Error # <module 'cv2.Error'>
import cv2.cuda as cuda # <module 'cv2.cuda'>
import cv2.detail as detail # <module 'cv2.detail'>
import cv2.dnn as dnn # <module 'cv2.dnn'>
import cv2.fisheye as fisheye # <module 'cv2.fisheye'>
import cv2.flann as flann # <module 'cv2.flann'>
import cv2.instr as instr # <module 'cv2.instr'>
import cv2.ipp as ipp # <module 'cv2.ipp'>
import cv2.ml as ml # <module 'cv2.ml'>
import cv2.ocl as ocl # <module 'cv2.ocl'>
import cv2.ogl as ogl # <module 'cv2.ogl'>
import cv2.samples as samples # <module 'cv2.samples'>
import cv2.utils as utils # <module 'cv2.utils'>
import cv2.videoio_registry as videoio_registry # <module 'cv2.videoio_registry'>
import cv2 as __cv2


from .object import object

class CascadeClassifier(object):
    # no doc
    def convert(self, oldcascade, newcascade): # real signature unknown; restored from __doc__
        """
        convert(oldcascade, newcascade) -> retval
        .
        """
        pass

    def detectMultiScale(self, image, scaleFactor=None, minNeighbors=None, flags=None, minSize=None, maxSize=None): # real signature unknown; restored from __doc__
        """
        detectMultiScale(image[, scaleFactor[, minNeighbors[, flags[, minSize[, maxSize]]]]]) -> objects
        .   @brief Detects objects of different sizes in the input image. The detected objects are returned as a list
        .   of rectangles.
        .   
        .   @param image Matrix of the type CV_8U containing an image where objects are detected.
        .   @param objects Vector of rectangles where each rectangle contains the detected object, the
        .   rectangles may be partially outside the original image.
        .   @param scaleFactor Parameter specifying how much the image size is reduced at each image scale.
        .   @param minNeighbors Parameter specifying how many neighbors each candidate rectangle should have
        .   to retain it.
        .   @param flags Parameter with the same meaning for an old cascade as in the function
        .   cvHaarDetectObjects. It is not used for a new cascade.
        .   @param minSize Minimum possible object size. Objects smaller than that are ignored.
        .   @param maxSize Maximum possible object size. Objects larger than that are ignored. If `maxSize == minSize` model is evaluated on single scale.
        .   
        .   The function is parallelized with the TBB library.
        .   
        .   @note
        .   -   (Python) A face detection example using cascade classifiers can be found at
        .   opencv_source_code/samples/python/facedetect.py
        """
        pass

    def detectMultiScale2(self, image, scaleFactor=None, minNeighbors=None, flags=None, minSize=None, maxSize=None): # real signature unknown; restored from __doc__
        """
        detectMultiScale2(image[, scaleFactor[, minNeighbors[, flags[, minSize[, maxSize]]]]]) -> objects, numDetections
        .   @overload
        .   @param image Matrix of the type CV_8U containing an image where objects are detected.
        .   @param objects Vector of rectangles where each rectangle contains the detected object, the
        .   rectangles may be partially outside the original image.
        .   @param numDetections Vector of detection numbers for the corresponding objects. An object's number
        .   of detections is the number of neighboring positively classified rectangles that were joined
        .   together to form the object.
        .   @param scaleFactor Parameter specifying how much the image size is reduced at each image scale.
        .   @param minNeighbors Parameter specifying how many neighbors each candidate rectangle should have
        .   to retain it.
        .   @param flags Parameter with the same meaning for an old cascade as in the function
        .   cvHaarDetectObjects. It is not used for a new cascade.
        .   @param minSize Minimum possible object size. Objects smaller than that are ignored.
        .   @param maxSize Maximum possible object size. Objects larger than that are ignored. If `maxSize == minSize` model is evaluated on single scale.
        """
        pass

    def detectMultiScale3(self, image, scaleFactor=None, minNeighbors=None, flags=None, minSize=None, maxSize=None, outputRejectLevels=None): # real signature unknown; restored from __doc__
        """
        detectMultiScale3(image[, scaleFactor[, minNeighbors[, flags[, minSize[, maxSize[, outputRejectLevels]]]]]]) -> objects, rejectLevels, levelWeights
        .   @overload
        .   This function allows you to retrieve the final stage decision certainty of classification.
        .   For this, one needs to set `outputRejectLevels` on true and provide the `rejectLevels` and `levelWeights` parameter.
        .   For each resulting detection, `levelWeights` will then contain the certainty of classification at the final stage.
        .   This value can then be used to separate strong from weaker classifications.
        .   
        .   A code sample on how to use it efficiently can be found below:
        .   @code
        .   Mat img;
        .   vector<double> weights;
        .   vector<int> levels;
        .   vector<Rect> detections;
        .   CascadeClassifier model("/path/to/your/model.xml");
        .   model.detectMultiScale(img, detections, levels, weights, 1.1, 3, 0, Size(), Size(), true);
        .   cerr << "Detection " << detections[0] << " with weight " << weights[0] << endl;
        .   @endcode
        """
        pass

    def empty(self): # real signature unknown; restored from __doc__
        """
        empty() -> retval
        .   @brief Checks whether the classifier has been loaded.
        """
        pass

    def getFeatureType(self): # real signature unknown; restored from __doc__
        """
        getFeatureType() -> retval
        .
        """
        pass

    def getOriginalWindowSize(self): # real signature unknown; restored from __doc__
        """
        getOriginalWindowSize() -> retval
        .
        """
        pass

    def isOldFormatCascade(self): # real signature unknown; restored from __doc__
        """
        isOldFormatCascade() -> retval
        .
        """
        pass

    def load(self, filename): # real signature unknown; restored from __doc__
        """
        load(filename) -> retval
        .   @brief Loads a classifier from a file.
        .   
        .   @param filename Name of the file from which the classifier is loaded. The file may contain an old
        .   HAAR classifier trained by the haartraining application or a new cascade classifier trained by the
        .   traincascade application.
        """
        pass

    def read(self, node): # real signature unknown; restored from __doc__
        """
        read(node) -> retval
        .   @brief Reads a classifier from a FileStorage node.
        .   
        .   @note The file may contain a new cascade classifier (trained traincascade application) only.
        """
        pass

    def __init__(self, *args, **kwargs): # real signature unknown
        pass

    @staticmethod # known case of __new__
    def __new__(*args, **kwargs): # real signature unknown
        """ Create and return a new object.  See help(type) for accurate signature. """
        pass

    def __repr__(self, *args, **kwargs): # real signature unknown
        """ Return repr(self). """
        pass

核心代码 

gray = cv.cvtColor(src, cv.COLOR_BGR2GRAY)
face_check = cv.CascadeClassifier("./haarcascades/haarcascade_frontalface_alt_tree.xml")  # 加载分类器
face = face_check.detectMultiScale(gray, 1.02, 1)  # 多尺寸检测人脸

实现人脸检测主要依赖于detectMultiScale()函数

def detectMultiScale(self, image, scaleFactor=None, minNeighbors=None, flags=None, minSize=None, maxSize=None)

代码实现

import cv2 as cv
import numpy as np


def  face_demo():
    gray = cv.cvtColor(src, cv.COLOR_BGR2GRAY)
    face_check = cv.CascadeClassifier("./haarcascades/haarcascade_frontalface_alt_tree.xml")  # 加载分类器
    face = face_check.detectMultiScale(gray, 1.02, 5)  # 多尺寸检测人脸
    for x, y, w, h in face:
        # 人脸区域框
        cv.rectangle(src, (x, y), (x + w, y + h), (0, 0, 255), 2)
    cv.imshow("result", src)


src = cv.imread("./../img/f.jpg")
cv.namedWindow("input img", cv.WINDOW_AUTOSIZE)
cv.imshow("input img", src)
face_demo()
cv.waitKey(0)
cv.destroyAllWindows()