图像二值化

二值图像：只有黑和白
0 代表黑***r> 1 其实是255 代表白色

图像二值化的方法

全局阈值
局部阈值

• OTSU
• Triangle
• 手动与自动
• 自适应阈值

相关代码

import cv2 as cv
import numpy as np


def threshold_demo(image):
    """ 全局阈值 :param image: :return: """
    # 首先图像灰度化处理
    gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
    # 函数参数 输入的图像 阈值 最大值 方法
    # 当方法使用了cv.THRESH_TRIANGLE 等自己寻找阈值时，函数中的参数2 指定的阈值就不起作用
    ret, binary = cv.threshold(gray, 0, 255, cv.THRESH_BINARY | cv.THRESH_OTSU)
    print("threshold value %s" % ret)
    cv.imshow("binary", binary)
    # 如果想要自己设定阈值 将参数2设定为自己设定的阈值 最常用
    ret, binary = cv.threshold(gray, 127, 255, cv.THRESH_BINARY)
    cv.imshow("127 binary", binary)
    # 反向
    ret, binary = cv.threshold(gray, 127, 255, cv.THRESH_BINARY_INV)
    cv.imshow("THRESH_BINARY_INV", binary)
    # cv.THRESH_TRUNC 截断 截断设定最大值为127
    ret, binary = cv.threshold(gray, 127, 255, cv.THRESH_TRUNC)
    cv.imshow("THRESH_TRUNC", binary)
    # 小于127的全部变为0
    ret, binary = cv.threshold(gray, 127, 255, cv.THRESH_TOZERO)
    cv.imshow("THRESH_TOZERO", binary)


def local_threshold(image):
    """ 局部阈值 可针对亮度不均匀情况 :param image: :return: """
    gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
    # 自适应阈值 参数:图像 最大值 方法(有俩个可选) 二值化 奇数 常量(均值)
    binary = cv.adaptiveThreshold(gray, 255, cv.ADAPTIVE_THRESH_GAUSSIAN_C, cv.THRESH_BINARY, 25, 10)
    cv.imshow("LOCAL binary", binary)


def custom_threshold(image):
    """ 自己计算图像的均值 设定为阈值 :param image: :return: """
    gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
    h, w = gray.shape[:2]
    # 变成一维的数组
    m = np.reshape(gray, [1, w * h])
    # 求所有的平均值
    mean = m.sum() / (w * h)
    print("mean : ", mean)
    ret, binary = cv.threshold(gray, mean, 255, cv.THRESH_BINARY)
    cv.imshow("custom_threshold binary", binary)


src = cv.imread("tree.jpg")
cv.namedWindow("input image", cv.WINDOW_AUTOSIZE)
cv.imshow("input image", src)
threshold_demo(src)
local_threshold(src)
custom_threshold(src)
cv.waitKey(0)

cv.destroyAllWindows()

结果展示