Python传统图像处理之皮肤区域检测详解

def skinMask_rgb(image):
    b, g, r = cv2.split(image)
    mask_uniformity = (r>95)*(g>40)*(b>20)* (np.max(image, axis=2) -np.min(image,axis=2)>15)* (abs(r-g)>15)*(r>g)*(r>b)
    mask_side = (r>220)*(g>210)*(b>170)*(abs(r-g)<=15)*(r>b)*(g>b)
    mask = mask_uniformity|mask_side
    skin = np.array(mask, np.uint8)*255
    ratio = np.sum(skin/255)/(image.shape[0]*image.shape[1])
 
    return skin, ratio

def skinMask_YCrCb(image):
    YCrCb = cv2.cvtColor(image, cv2.COLOR_BGR2YCR_CB) #转换至YCrCb空间
    (y,cr,cb) = cv2.split(YCrCb) #拆分出Y,Cr,Cb值
    skin = np.zeros(cr.shape, dtype = np.uint8)
    (x,y) = cr.shape
    
    cr_mask1 = cr >= 133
    cr_mask2 = cr <= 173
    cr_mask = cr_mask1 * cr_mask2
 
    cb_mask1 = cb >= 77
    cb_mask2 = cb <= 127
    cb_mask = cb_mask1 * cb_mask2
    mask = cr_mask * cb_mask
    skin[mask] = 255
 
    ratio = np.sum(skin/255)/(x*y)
    return skin, ratio

def skinMask_YCrCb_otsu(image):
    YCrCb = cv2.cvtColor(image, cv2.COLOR_BGR2YCR_CB) #转换至YCrCb空间
    (y,cr,cb) = cv2.split(YCrCb) #拆分出Y,Cr,Cb值
    
    ret, skin = cv2.threshold(cr, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)
 
    ratio = np.sum(skin/255)/(image.shape[0]*image.shape[1])
    return skin, ratio

def skinMask_hsv(image):
    hsv = cv2.cvtColor(image, cv2.COLOR_BGR2HSV)
    h, s, v = cv2.split(hsv)
 
    mask = (h>7)*(h<20)*(s>28)*(v>50)
    skin = np.array(mask, np.uint8)*255
    ratio = np.sum(skin/255)/(image.shape[0]*image.shape[1])
    return skin, ratio

def skinMask_ellipse(image):
    YCrCb = cv2.cvtColor(image, cv2.COLOR_BGR2YCR_CB) #转换至YCrCb空间
    (y,cr,cb) = cv2.split(YCrCb) #拆分出Y,Cr,Cb值
 
    skinCrCbHist = np.zeros((256,256),np.uint8)
    skinCrCbHist = cv2.ellipse(skinCrCbHist, (113, 155), (23, 15), 43.0, 0.0, 360.0, (255, 255, 255), -1)
    cv2.imwrite("ww.jpg", skinCrCbHist)
 
    skin = cv2.remap(skinCrCbHist, cb.astype(np.float32), cr.astype(np.float32), interpolation=cv2.INTER_LINEAR)
    ratio = np.sum(skin/255)/(image.shape[0]*image.shape[1])
    return skin, ratio

def skinMask_hist_backproject(target, roi_image):
    #roi图片，就想要找的的图片
    hsv = cv2.cvtColor(roi_image,cv2.COLOR_BGR2HSV)
 
    #目标搜索图片
    hsvt = cv2.cvtColor(target,cv2.COLOR_BGR2HSV)
 
    #计算目标直方图
    roihist = cv2.calcHist([hsv],[0,1],None,[180,256],[0,180,0,256])
    #归一化，参数为原图像和输出图像，归一化后值全部在2到255范围
    cv2.normalize(roihist,roihist,0,255,cv2.NORM_MINMAX)
    dst = cv2.calcBackProject([hsvt],[0,1],roihist,[0,180,0,256],1)
 
 
    #卷积连接分散的点
    disc = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(5,5))
    dst = cv2.filter2D(dst,-1,disc)
 
    ret,skin = cv2.threshold(dst,50,255,0)
 
    ratio = np.sum(skin/255)/(target.shape[0]*target.shape[1])
return skin, ratio