opencv追踪眼球.rar

''' 作者：树莓 原理：通过面部识别和眼部模型识别找出面部和眼部 再把两个眼睛分别截取出来（剔除眉毛干扰） 做二值化处理，寻找轮廓，对眼珠进行圆形拟合 效果其实一般般 ''' #需要安装上面的两个库 import numpy as np import cv2 def nothing(args): pass #两个分类器目录一定要选对，就是文件夹里面的两个文件 # 人脸识别分类器 faceCascade = cv2.CascadeClassifier(r"C:\Users\TSK\Desktop\opencv_find_eye/haarcascade_frontalface_default.xml") # 识别眼睛的分类器 eyeCascade = cv2.CascadeClassifier(r'C:\Users\TSK\Desktop\opencv_find_eye/haarcascade_eye.xml') # 开启摄像头 cap = cv2.VideoCapture(0) ok=True cv2.namedWindow('eye1') cv2.namedWindow('eye2') #这两个滑块用来调整阈值 cv2.createTrackbar("th","eye1",20,255,nothing) cv2.createTrackbar("th","eye2",20,255,nothing) while ok: # 读取摄像头中的图像，ok为是否读取成功的判断参数 ok, img = cap.read() img2 =img # 转换成灰度图像 gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # 人脸检测 faces = faceCascade.detectMultiScale( gray, scaleFactor=1.2, minNeighbors=5, minSize=(32, 32) ) # 在检测人脸的基础上检测眼睛 for (x, y, w, h) in faces: fac_gray = gray[y: (y+h), x: (x+w)] result = [] eyes = eyeCascade.detectMultiScale(fac_gray, 1.3, 8,cv2.CASCADE_SCALE_IMAGE,(40,40),(80,80)) # 眼睛坐标的换算，将相对位置换成绝对位置 for (ex, ey, ew, eh) in eyes: #print(eyes[0]) result.append((x+ex, y+ey, ew, eh)) # 画矩形 for (x, y, w, h) in faces: cv2.rectangle(img, (x, y), (x+w, y+h), (255, 0, 0), 2) try : for (ex, ey, ew, eh) in result: cv2.rectangle(img, (ex, ey), (ex+ew, ey+eh), (0, 255, 0), 2)#框出眼睛 if result[0][0] >result[1][0]: #判断是左眼还是右眼 img_eye_l = cv2.resize(img2[result[0][1]:result[0][1]+result[0][3], result[0][0]:result[0][0]+result[0][2]], (300, 300)) img_eye_r = cv2.resize(img2[result[1][1]:result[1][1]+result[1][3], result[1][0]:result[1][0]+result[1][2]], (300, 300)) if result[0][0] <result[1][0]: img_eye_l = cv2.resize(img2[result[1][1]:result[1][1]+result[1][3], result[1][0]:result[1][0]+result[1][2]], (300, 300)) img_eye_r = cv2.resize(img2[result[0][1]:result[0][1]+result[0][3], result[0][0]:result[0][0]+result[0][2]], (300, 300)) img_eye_l = img_eye_l[60:240,10:290] #复制眼睛照片 img_eye_r = img_eye_r[60:240,10:290] img_eye_l_gray = cv2.cvtColor(img_eye_l,cv2.COLOR_RGB2GRAY) #转换为灰度图 img_eye_r_gray = cv2.cvtColor(img_eye_r,cv2.COLOR_RGB2GRAY) X1 = cv2.getTrackbarPos("th","eye1") #获取滑块的值 X2 = cv2.getTrackbarPos("th","eye2") ret1,adaptive_l = cv2.threshold(img_eye_l_gray,X1,255,cv2.THRESH_BINARY_INV) #普通二值化处理 ret2,adaptive_r = cv2.threshold(img_eye_r_gray,X2,255,cv2.THRESH_BINARY_INV) #adaptive_l = cv2.adaptiveThreshold(img_eye_l_gray,255,cv2.ADAPTIVE_THRESH_MEAN_C,cv2.THRESH_BINARY,81,2) #自适应二值化 效果并不好 #adaptive_r = cv2.adaptiveThreshold(img_eye_r_gray,255,cv2.ADAPTIVE_THRESH_MEAN_C,cv2.THRESH_BINARY,81,2) mask_l = cv2.erode(adaptive_l, None, iterations=2) #收缩 mask_r = cv2.erode(adaptive_r, None, iterations=2) mask_l = cv2.dilate(mask_l, None, iterations=2) #膨胀 收缩膨胀主要是为了过滤噪点 mask_r = cv2.dilate(mask_r, None, iterations=2) cnts_l = cv2.findContours(mask_l.copy(),cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)[-2] #寻找轮廓 简单轮廓 cnts_r = cv2.findContours(mask_r.copy(),cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)[-2] if len(cnts_l) >0: #圆形拟合 c = max(cnts_l, key=cv2.contourArea) ((x, y), radius) = cv2.minEnclosingCircle(c) M = cv2.moments(c) center = (int(M["m10"] / M["m00"]), int(M["m01"] / M["m00"])) if radius > 3: cv2.circle(img_eye_l, (int(x), int(y)), int(radius), (255, 0, 0), 2) cv2.circle(img_eye_l, center, 5, (0, 0, 255), -1) if len(cnts_r) >0: #圆形拟合 c = max(cnts_r, key=cv2.contourArea) ((x, y), radius) = cv2.minEnclosingCircle(c) M = cv2.moments(c) center = (int(M["m10"] / M["m00"]), int(M["m01"] / M["m00"])) if radius > 3: cv2.circle(img_eye_r, (int(x), int(y)), int(radius), (255, 0, 0), 2) cv2.circle(img_eye_r, center, 5, (0, 0, 255), -1) cv2.imshow("eye1",img_eye_l) cv2.imshow("eye2",img_eye_r) #cv2.imshow("eye_l",img_eye_l_gray) #cv2.imshow("eye_r",img_eye_r_gray) #cv2.imshow("eye_l",adaptive_l) #cv2.imshow("eye_r",adaptive_r) except: pass cv2.imshow('video', img) k = cv2.waitKey(1) if k == 27: # press 'ESC' to quit break cap.release() cv2.destroyAllWindows()