python opencv 按照hsv筛选过滤背景图片

---

• 20190925

---

1. 博客背景

   高分辨率图像为了方便处理，经常会有切分为小图的应用场景。在切分过程中总会有些小图中背景图像占很大一部分，需要再筛选过滤掉。

2. 代码实现

   此处只展示计算白色背景占比。

   import cv2
   import numpy as np
   from skimage import io
   import matplotlib.pyplot as plt 
   import skimage.morphology as sm
   # import skimage.filters.rank as sfr
   
   def patch_white_percent(img):
       hsv=cv2.cvtColor(img,cv2.COLOR_RGB2HSV)
       # red_mask=cv2.inRange(hsv,np.array([156,43,46]),np.array([180,255,255]))
       # purple_mask=cv2.inRange(hsv,np.array([125,43,46]),np.array([155,255,255]))
       # blue_mask=cv2.inRange(hsv,np.array([100,43,46]),np.array([124,255,255]))
       # cyan_mask=cv2.inRange(hsv,np.array([78,43,46]),np.array([99,255,255]))
       # green_mask=cv2.inRange(hsv,np.array([35,43,46]),np.array([77,255,255]))
       # yellow_mask=cv2.inRange(hsv,np.array([26,43,46]),np.array([34,255,255]))
       # orange_mask=cv2.inRange(hsv,np.array([11,43,46]),np.array([25,255,255]))
       # # gray_mask=cv2.inRange(hsv,np.array([0,0,46]),np.array([180,43,220]))
       # # black_mask=cv2.inRange(hsv,np.array([0,0,0]),np.array([180,255,46]))
       # mask1=red_mask+purple_mask+blue_mask+cyan_mask+green_mask+yellow_mask+orange_mask+gray_mask+black_mask
       # mask1_temp = np.where(mask1>1,1,0)
   
       white_mask = cv2.inRange(hsv , np.array([0,0,221]) , np.array([180,30,255]))
       
   	#### 因为白色是不需要的颜色，所以做了先开后闭的处理
       dst = sm.opening(white_mask,sm.disk(5))
       # dst = sfr.median(dst, sm.disk(3)) #中值滤波去小点（此处作用不大，可以省略）
       white = sm.closing(dst,sm.disk(5))
   
       plt.subplot(221)
       plt.title('img')
       plt.imshow(org_thumnail)
       plt.subplot(222)
       plt.title('white_mask')
       plt.imshow(white_mask)
       plt.subplot(223)
       plt.title('white_end')
       plt.imshow(white)
       plt.show()
       
   	#### 计算白色背景占比
       mask1_temp = np.where(white>1,1,0)
       percent_temp = np.sum(mask1_temp)/(mask1_temp.shape[0]*mask1_temp.shape[1])
       return percent_temp
   
   img = r"C:\Data\20190917_ndpi\20190925171451342_01_HER2_2816_4352.png"
   img = io.imread(img)
   
   percent_temp = patch_white_percent(img)
   print('#### percent_temp',percent_temp)
   

3. 输入图像与结果展示
   
   

   #### percent_temp 0.2086181640625