图像的基本运算——阈值化处理
图像阈值化处理有五种方式:二进制阈值化、反二进制阈值化、截断阈值化、反阈值化为0、阈值化为0。如图所示,上面的图为原始图像像素分布,下面五种分别是阈值化处理后的像素分布图。
-
二进制阈值化
#假设阈值为127 r, b = cv2.threshold(a,127,255,cv2.THRESH_BINARY)#一定要用灰度图像
-
反二进制阈值化
#假设阈值为127 r, c = cv2.threshold(a,127,255,cv2.THRESH_BINARY_INV)
-
截断阈值化
#假设阈值为127 r, c = cv2.threshold(a,127,255,cv2.THRESH_TRUNC)
-
反阈值化为0
#假设阈值为127
r, b = cv2.threshold(a,127,255,THRESH_TOZERO_INV)
- 阈值化为0
#假设阈值为127
r, b = cv2.threshold(a,127,255,THRESH_TOZERO_INV)
示例:
import cv2
a = cv2.imread("image\\lena512.bmp",cv2.IMREAD_UNCHANGED)
#原始图像的类型,必须是8位单通道图像,彩色图像无意义
r, b = cv2.threshold(a,127,255,cv2.THRESH_TOZERO_INV)#一定要用灰度图像
r, c = cv2.threshold(a,127,255,cv2.THRESH_TRUNC)
cv2.imshow("a",a)
cv2.imshow("b",b)
cv2.imshow("c",c)
cv2.waitKey()
cv2.destroyAllWindows()
以上阈值化的方法是选定一个阈值来处理,阈值化处理我们还可以通过一个阈值范围来调整图像,这里给大家分享一个项目案例:
import cv2
import numpy as np
def empty(a):
pass
a = cv2.imread("imag",cv2.IMREAD_UNCHANGED)
b = cv2.cvtColor(a,cv2.COLOR_BGR2GRAY)
cv2.namedWindow("tracking")
cv2.createTrackbar("l_T","tracking",0,255,empty)
cv2.createTrackbar("u_T","tracking",255,255,empty)
while True:
l_T = cv2.getTrackbarPos("l_T", "tracking")
u_T = cv2.getTrackbarPos("u_T", "tracking")
_,mask=cv2.threshold(b,l_T,u_T,cv2.THRESH_BINARY)
cv2.imshow("orignial",b)
cv2.imshow('mask',mask)
key = cv2.waitKey(1) # Esc键
if key == 27:
break
cv2.destroyAllWindows()
效果展示: