環境準備

編輯器 pycharm
python 版本3.7x

pip3 install opencv-python
pip3 install opencv-contrib-python

基本要求：
熟練掌握 python3.x
懂得基本Pycahrm操作以及圖片基本知識

資源讀取

讀取圖片

import cv2.cv2 as cv

Image = cv.imread("Image/face1.jpg")
cv.imshow("Image",Image)
cv.waitKey(0)
cv.destroyAllWindows()

讀取視頻

import cv2

cap = cv2.VideoCapture(0)
# 0 表示讀取攝像頭
# 加入具體資源路徑就表示直接讀取路徑對應的媒體資源

while True:
    flag,img = cap.read()
    cv2.imshow("Vido",img)
    if(cv2.waitKey(1)==ord("q")):
        #cv2.waitKey(0) 表示死回圈等待
        #cv2.waitKey(time)表示等待time ms 記錄用戶的輸入獲取對應的ASCII值
        #這里表示等待一ms并且如果用戶輸入 q 就退出
        break

資源操作

修改圖片

繪制圖形

這個在我們的圖片框選當中是很有必要的

#繪制直線
cv2.line( img , (0,0),(511,511),( 255,0,0),5)
#繪制矩形
cv2.rectangle(img,(x,y,x+w,y+h),color=(0,0,255),thickness=1) # 方框顏色，粗細 -1表示實心
#繪制圓形
cv2.circle(img,center=(x+w,y+h),radius=100,color=(255,0,0),thickness=5)
#寫字
font = cv2.FONT_HERSHEY_SIMPLEX
cv2.putText(img , ' OpencV',(10,500)， font，4,(255,255,255),2,cv.LINE_AA)
#在哪顯示，顯示字體，顯示位置（開始），字體，大小，顏色，線條粗細，什么線條（實線，虛線...）

尺寸修改

Image_resize = cv.resize(Image,dsize=(200,200))

灰度處理圖片

gary = cv.cvtColor(img,cv.COLOR_BGR2GRAY)

事實上

Image = cv.imread("Image/face1.jpg",0)

也可以，但是我們讀取攝像頭的時候是吧，有要處理的，
例如：

import cv2

cap = cv2.VideoCapture(0)
cap.set(3,640)
cap.set(4,480)

while True:
    flag,img = cap.read()
    img = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)

    cv2.imshow("Vido",img)
    if(cv2.waitKey(1)==ord("q")):
        #cv2.waitKey(0) 表示死回圈等待
        #cv2.waitKey(time)表示等待time ms 記錄用戶的輸入獲取對應的ASCII值
        #這里表示等待一ms并且如果用戶輸入 q 就退出
        break

圖片保存

cv2.imwrite(filename,imgage)

使用模型

import cv2.cv2 as cv
import cv2.data as data

#讀取影像
img = cv.imread('Image/face1.jpg')
gary = cv.cvtColor(img,cv.COLOR_BGR2GRAY)
face_detect = cv.CascadeClassifier(data.haarcascades+"haarcascade_frontalface_alt2.xml")
face = face_detect.detectMultiScale(gary)
for x,y,w,h in face:
    cv.rectangle(img,(x,y),(x+w,y+h),color=(0,0,255),thickness=2)
cv.imshow('result',img)

cv.waitKey(0)
#釋放記憶體
cv.destroyAllWindows()

CascadeClassifier()就是加載

例如：

import cv2.cv2 as cv
import cv2.data as data
cap = cv.VideoCapture(0)
cap.set(3,640)
cap.set(4,480)

while True:
    flag,img = cap.read()
    gary = cv.cvtColor(img,cv.COLOR_BGR2GRAY)

    face_detect = cv.CascadeClassifier(data.haarcascades + "haarcascade_frontalface_alt2.xml")
    face = face_detect.detectMultiScale(gary)
    for x, y, w, h in face:
        cv.rectangle(img, (x, y), (x + w, y + h), color=(0, 0, 255), thickness=2)

    cv.imshow("Vido",img)
    if(cv.waitKey(1)==ord("q")):
        #cv2.waitKey(0) 表示死回圈等待
        #cv2.waitKey(time)表示等待time ms 記錄用戶的輸入獲取對應的ASCII值
        #這里表示等待一ms并且如果用戶輸入 q 就退出
        break

cap.release()
#釋放記憶體
cv.destroyAllWindows()

模型訓練

recognizer=cv.face.LBPHFaceRecognizer_create()
recognizer.train(faces,np.array(ids))
ids, confidence = recogizer.predict(gray[y:y + h, x:x + w])

加載預測模型

這個直接看例子吧：

import os
import cv2.cv2 as cv
import numpy as np
import cv2.data as data
def getImageIds(path):
    #函式作用是提取人臉然后回傳人物的人臉和id
    faceseare=[] # 保存檢測出的人臉
    ids=[] #
    imagePaths=[os.path.join(path,f) for f in os.listdir(path)]
    #人臉檢測
    face_detector = cv.CascadeClassifier(data.haarcascades+"haarcascade_frontalface_default.xml")
    if imagePaths:
        print('訓練圖片為：',imagePaths)
    else:
        print("請先錄入人臉")
        return

    for imagePath in imagePaths:
        #二值化處理
        img = cv.imread(imagePath)
        img=cv.cvtColor(img,cv.COLOR_BGR2GRAY)
        img_numpy=np.array(img,'uint8')#獲取圖片矩陣

        faces = face_detector.detectMultiScale(img_numpy)

        id = int(os.path.split(imagePath)[1].split('.')[0])

        for x,y,w,h in faces:
            ids.append(id)
            faceseare.append(img_numpy[y:y+h,x:x+w])
        print('已獲取id:', id)

    return faceseare,ids

if __name__ == '__main__':
    #圖片路徑
    path='Image/InPutImg'
    #獲取影像陣列和id標簽陣列和姓名
    faces,ids=getImageIds(path)
    #獲取訓練物件
    recognizer=cv.face.LBPHFaceRecognizer_create()
    recognizer.train(faces,np.array(ids)) #把對應的人臉和id聯系起來訓練
    #保存訓練檔案
    model_save = "trainer/"
    if not os.path.exists(model_save):
        os.makedirs(model_save)
    recognizer.write('trainer/trainer.yml')

完整的例子如下：
https://blog.csdn.net/futerox/article/details/120685898

訓練模型加載

這個也是例子
使用
recogizer=cv.face.LBPHFaceRecognizer_create()#加載訓練資料集檔案
recogizer.read(‘trainer/trainer.yml’)

import cv2.cv2 as cv
import os
import cv2.data as data

recogizer=cv.face.LBPHFaceRecognizer_create()#加載訓練資料集檔案
recogizer.read('trainer/trainer.yml')
names=[]
warningtime = 0

#準備識別的圖片
def face_detect_demo(img):
    global warningtime
    gray=cv.cvtColor(img,cv.COLOR_BGR2GRAY)
    face_detector=cv.CascadeClassifier(data.haarcascades+"haarcascade_frontalface_default.xm")
    face=face_detector.detectMultiScale(gray)
    for x,y,w,h in face:
        cv.rectangle(img,(x,y),(x+w,y+h),color=(0,0,255),thickness=2)
        cv.circle(img,center=(x+w//2,y+h//2),radius=w//2,color=(0,255,0),thickness=1)

        # 人臉識別
        ids, confidence = recogizer.predict(gray[y:y + h, x:x + w])

        if confidence > 80: #評分越大可信度越低
            warningtime += 1
            if warningtime > 100:
               warningtime = 0
               print("未識別出此人") #這塊的話其實可以再搞一套對應的懲罰機制
            cv.putText(img, 'unkonw', (x + 10, y - 10), cv.FONT_HERSHEY_SIMPLEX, 0.75, (0, 255, 0), 1)
        else:
        #這里也對應一套識別后的機制
            cv.putText(img,str(names[ids-1]), (x + 10, y - 10), cv.FONT_HERSHEY_SIMPLEX, 0.75, (0, 255, 0), 1)
    cv.imshow('result',img)


def get_name(names):
    path = 'Image/InPutImg/'

    imagePaths=[os.path.join(path,f) for f in os.listdir(path)]
    for imagePath in imagePaths:
       name = str(os.path.split(imagePath)[1].split('.',2)[1])
       names.append(name)


cap=cv.VideoCapture(0)
get_name(names)
while True:
    flag,frame=cap.read()
    if not flag:
        break
    face_detect_demo(frame)
    if ord(' ') == cv.waitKey(10): #按下空格關了
        break
cap.release()
cv.destroyAllWindows()

手勢識別

純opencv版本

這個的話有點超綱了，在本篇博文，其實寫這篇博文的目的還是翻筆記的時候找到了這篇筆記，順便整理一下，其實就是那篇
分分鐘自制人臉識別（如何快速識別心儀的小姐姐~）
的邊角料
但是，突然想到最近那個手勢識別比較火，那就來一下唄，識別你的手勢是哪個數字，識別1~5
這個原理如下：

1.選擇手勢區域，我們這個比較那啥需要選定一個區域來放置我們的手勢
2. 進行高斯濾波
3.之后我們就能夠通過找到手的輪廓，之后找到凹凸位置，那么這樣一來就能夠判斷你的手勢是幾

那么這個局限性也就很明顯了，我們是通過凹凸點來確定你的手勢是數字幾的，這個就有點那啥了，只能說能用，

import cv2
import numpy as np
import math

cap = cv2.VideoCapture(0)

while (cap.isOpened()):
    ret, frame = cap.read()  # 讀取攝像頭每幀圖片

    frame = cv2.flip(frame, 1)
    kernel = np.ones((2, 2), np.uint8)
    roi = frame[100:600, 100:600]  # 選取圖片中固定位置作為手勢輸入

    cv2.rectangle(frame, (100, 100), (600, 600), (0, 0, 255), 0)  # 用紅線畫出手勢識別框
    # 基于hsv的膚色檢測
    hsv = cv2.cvtColor(roi, cv2.COLOR_BGR2HSV)
    lower_skin = np.array([0, 28, 70], dtype=np.uint8)
    upper_skin = np.array([20, 255, 255], dtype=np.uint8)

    # 進行高斯濾波
    mask = cv2.inRange(hsv, lower_skin, upper_skin)
    mask = cv2.dilate(mask, kernel, iterations=4)
    mask = cv2.GaussianBlur(mask, (5, 5), 100)

    # 找出輪廓
    contours, h = cv2.findContours(mask, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
    cnt = max(contours, key=lambda x: cv2.contourArea(x))
    epsilon = 0.0005 * cv2.arcLength(cnt, True)
    approx = cv2.approxPolyDP(cnt, epsilon, True)
    hull = cv2.convexHull(cnt)
    areahull = cv2.contourArea(hull)
    areacnt = cv2.contourArea(cnt)
    arearatio = ((areahull - areacnt) / areacnt) * 100

    # 求出凹凸點
    hull = cv2.convexHull(approx, returnPoints=False)
    defects = cv2.convexityDefects(approx, hull)

    # 定義凹凸點個數初始值為0
    l = 0

    for i in range(defects.shape[0]):
        s, e, f, d, = defects[i, 0]
        start = tuple(approx[s][0])
        end = tuple(approx[e][0])
        far = tuple(approx[f][0])
        pt = (100, 100)

        a = math.sqrt((end[0] - start[0]) ** 2 + (end[1] - start[1]) ** 2)
        b = math.sqrt((far[0] - start[0]) ** 2 + (far[1] - start[1]) ** 2)
        c = math.sqrt((end[0] - far[0]) ** 2 + (end[1] - far[1]) ** 2)
        s = (a + b + c) / 2
        ar = math.sqrt(s * (s - a) * (s - b) * (s - c))

        # 手指間角度求取
        angle = math.acos((b ** 2 + c ** 2 - a ** 2) / (2 * b * c)) * 57

        if angle <= 90 and d > 20:
            l += 1
            cv2.circle(roi, far, 3, [255, 0, 0], -1)
        cv2.line(roi, start, end, [0, 255, 0], 2)  # 畫出包絡線
    l += 1
    font = cv2.FONT_HERSHEY_SIMPLEX

    # 條件判斷，知道手勢后想實作的功能
    if l == 1:
        if areacnt < 2000:
            cv2.putText(frame, "put hand in the window", (0, 50), font, 2, (0, 0, 255), 3, cv2.LINE_AA)
        else:
            if arearatio < 12:
                cv2.putText(frame, '0', (0, 50), font, 2, (0, 0, 255), 3, cv2.LINE_AA)
            elif arearatio < 17.5:
                cv2.putText(frame, "1", (0, 50), font, 2, (0, 0, 255), 3, cv2.LINE_AA)
            else:
                cv2.putText(frame, '1', (0, 50), font, 2, (0, 0, 255), 3, cv2.LINE_AA)
    elif l == 2:
        cv2.putText(frame, '2', (0, 50), font, 2, (0, 0, 255), 3, cv2.LINE_AA)
    elif l == 3:
        if arearatio < 27:
            cv2.putText(frame, '3', (0, 50), font, 2, (0, 0, 255), 3, cv2.LINE_AA)
        else:
            cv2.putText(frame, '3', (0, 50), font, 2, (0, 0, 255), 3, cv2.LINE_AA)
    elif l == 4:
        cv2.putText(frame, '4', (0, 50), font, 2, (0, 0, 255), 3, cv2.LINE_AA)
    elif l == 5:
        cv2.putText(frame, '5', (0, 50), font, 2, (0, 0, 255), 3, cv2.LINE_AA)
    cv2.imshow('frame', frame)
    cv2.imshow('mask', mask)

    key = cv2.waitKey(25) & 0xFF
    if key == ord('q'):  # 鍵盤q鍵退出
        break

cv2.destroyAllWindows()
cap.release()

此外的話還有一個版本的，這個是我在找資料的時候發現的，他這個是使用了 mediapipe 這個玩意來做的，結合opencv

mediapipe版本

參考這幾個人的博客：
https://blog.csdn.net/weixin_43654363/article/details/120809464
https://blog.csdn.net/qq_43550173/article/details/116273477
以及這個視頻(滿屏咖喱味)
https://www.bilibili.com/video/BV1V34y1Q7ka?from=search&seid=13134349602411946620&spm_id_from=333.337.0.0

這里的代碼是CV除錯之后的，原來的別人寫的代碼有點問題，我這邊除錯了，

兩個腳本

HandTrackingModule.py

import cv2
import mediapipe as mp

class HandDetector:
    """
    使用mediapipe庫查找手，匯出地標像素格式，添加了額外的功能，
    如查找方式，許多手指向上或兩個手指之間的距離，而且提供找到的手的邊界框資訊，
    """
    def __init__(self, mode=False, maxHands=2,comPlexity=1, detectionCon=0.5, minTrackCon=0.5):
        """
        :param mode: 在靜態模式下，對每個影像進行檢測
        :param maxHands: 要檢測的最大手數
        :param detectionCon: 最小檢測置信度
        :param minTrackCon: 最小跟蹤置信度
        """
        self.mode = mode
        self.maxHands = maxHands
        self.detectionCon = detectionCon
        self.minTrackCon = minTrackCon
        self.comPlexity = comPlexity

        self.mpHands = mp.solutions.hands
        self.hands = self.mpHands.Hands(self.mode, self.maxHands,self.comPlexity,
                                        self.detectionCon, self.minTrackCon)
        self.mpDraw = mp.solutions.drawing_utils
        self.tipIds = [4, 8, 12, 16, 20]
        self.fingers = []
        self.lmList = []

    def findHands(self, img, draw=True):
        """
        從影像(BRG)中找到手部，
        :param img: 用于查找手的影像，
        :param draw: 在影像上繪制輸出的標志，
        :return: 帶或不帶圖形的影像
        """
        imgRGB = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) # 將傳入的影像由BGR模式轉標準的Opencv模式——RGB模式，
        self.results = self.hands.process(imgRGB)

        if self.results.multi_hand_landmarks:
            for handLms in self.results.multi_hand_landmarks:
                if draw:
                    self.mpDraw.draw_landmarks(img, handLms,
                                               self.mpHands.HAND_CONNECTIONS)
        return img

    def findPosition(self, img, handNo=0, draw=True):
        """
        查找單手的地標并將其放入串列中像素格式，還可以回傳手部周圍的邊界框，
        :param img: 要查找的主影像
        :param handNo: 如果檢測到多只手，則為手部id
        :param draw: 在影像上繪制輸出的標志，(默認繪制矩形框)
        :return: 像素格式的手部關節位置串列；手部邊界框
        """

        xList = []
        yList = []
        bbox = []
        bboxInfo =[]
        self.lmList = []
        if self.results.multi_hand_landmarks:
            myHand = self.results.multi_hand_landmarks[handNo]
            for id, lm in enumerate(myHand.landmark):
                h, w, c = img.shape
                px, py = int(lm.x * w), int(lm.y * h)
                xList.append(px)
                yList.append(py)
                self.lmList.append([px, py])
                if draw:
                    cv2.circle(img, (px, py), 5, (255, 0, 255), cv2.FILLED)
            xmin, xmax = min(xList), max(xList)
            ymin, ymax = min(yList), max(yList)
            boxW, boxH = xmax - xmin, ymax - ymin
            bbox = xmin, ymin, boxW, boxH
            cx, cy = bbox[0] + (bbox[2] // 2), \
                     bbox[1] + (bbox[3] // 2)
            bboxInfo = {"id": id, "bbox": bbox,"center": (cx, cy)}

            if draw:
                cv2.rectangle(img, (bbox[0] - 20, bbox[1] - 20),
                              (bbox[0] + bbox[2] + 20, bbox[1] + bbox[3] + 20),
                              (0, 255, 0), 2)

        return self.lmList, bboxInfo

    def fingersUp(self):
        """
        查找串列中打開并回傳的手指數，會分別考慮左手和右手
        ：return：豎起手指的陣列(串列)，陣列長度為5，
        其中，由大拇指開始數，立起標為1，放下為0，
        """
        if self.results.multi_hand_landmarks:
            myHandType = self.handType()
            fingers = []
            # Thumb
            if myHandType == "Right":
                if self.lmList[self.tipIds[0]][0] > self.lmList[self.tipIds[0] - 1][0]:
                    fingers.append(1)
                else:
                    fingers.append(0)
            else:
                if self.lmList[self.tipIds[0]][0] < self.lmList[self.tipIds[0] - 1][0]:
                    fingers.append(1)
                else:
                    fingers.append(0)

            # 4 Fingers
            for id in range(1, 5):
                if self.lmList[self.tipIds[id]][1] < self.lmList[self.tipIds[id] - 2][1]:
                    fingers.append(1)
                else:
                    fingers.append(0)
        return fingers

    def handType(self):
        """
        檢查傳入的手部是左還是右
        ：return: "Right" 或 "Left"
        """
        if self.results.multi_hand_landmarks:
            if self.lmList[17][0] < self.lmList[5][0]:
                return "Right"
            else:
                return "Left"

Main.py

import cv2
from Head.HandTrackingModule import HandDetector
class Main:
    def __init__(self):
        self.camera = cv2.VideoCapture(0,cv2.CAP_DSHOW)
        self.camera.set(3, 640)
        self.camera.set(4, 480)

    def Gesture_recognition(self):
        while True:
            self.detector = HandDetector()
            frame, img = self.camera.read()
            img = self.detector.findHands(img)
            lmList, bbox = self.detector.findPosition(img)

            if lmList:
                x_1, y_1 = bbox["bbox"][0], bbox["bbox"][1]
                x1, x2, x3, x4, x5 = self.detector.fingersUp()

                if (x2 == 1 and x3 == 1) and (x4 == 0 and x5 == 0 and x1 == 0):
                    cv2.putText(img, "2", (x_1, y_1), cv2.FONT_HERSHEY_PLAIN, 3,
                                (0, 0, 255), 3)
                elif (x2 == 1 and x3 == 1 and x4 == 1) and (x1 == 0 and x5 == 0):
                    cv2.putText(img, "3", (x_1, y_1), cv2.FONT_HERSHEY_PLAIN, 3,
                                (0, 0, 255), 3)
                elif (x2 == 1 and x3 == 1 and x4 == 1 and x5 == 1) and (x1 == 0):
                    cv2.putText(img, "4", (x_1, y_1), cv2.FONT_HERSHEY_PLAIN, 3,
                                (0, 0, 255), 3)
                elif x1 == 1 and x2 == 1 and x3 == 1 and x4 == 1 and x5 == 1:
                    cv2.putText(img, "5", (x_1, y_1), cv2.FONT_HERSHEY_PLAIN, 3,
                                (0, 0, 255), 3)
                elif x2 == 1 and (x1 == 0, x3 == 0, x4 == 0, x5 == 0):
                    cv2.putText(img, "1", (x_1, y_1), cv2.FONT_HERSHEY_PLAIN, 3,
                                (0, 0, 255), 3)
                elif x1 and (x2 == 0, x3 == 0, x4 == 0, x5 == 0):
                    cv2.putText(img, "GOOD!", (x_1, y_1), cv2.FONT_HERSHEY_PLAIN, 3,
                                (0, 0, 255), 3)
            cv2.imshow("camera", img)
            if cv2.getWindowProperty('camera', cv2.WND_PROP_VISIBLE) < 1:
                break
            if cv2.waitKey(1)==ord("q"):
                break

if __name__ == '__main__':
    Solution = Main()
    Solution.Gesture_recognition()

效果