【Python】OpenCV库学习笔记（六）

1. 图像二值化

1.1 二值图像（Binary Image）

（1）二值图像的概念

• 一张图像只有黑白两个颜色
• 下面图像是就是一个二值图像，其中的1代表黑色，其中的0代表白色

（2）二值图像的存储

• 使用四叉树存储平面图像

1.2 图像二值化方法

图像二值化的方法：

1. 全局阈值
2. 局部阈值

1.3 OpenCV相关API

1.3.1 OTSU

import cv2 as cv
import numpy

def threshold_demo(image):
    gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
    ret, binary = cv.threshold(gray, 0, 255, cv.THRESH_BINARY | cv.THRESH_OTSU)
    print("threshold value %s" % ret)
    cv.imshow("Binary IMage", binary)

image = cv.resize(cv.imread("01.jpg"), (300,400))
cv.imshow("image", image)
threshold_demo(image)
cv.waitKey(0)
cv.destroyAllWindows()

threshold value 171.0

1.3.2 Triangle

import cv2 as cv
import numpy


def threshold_demo(image):
    gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
    ret, binary = cv.threshold(gray, 0, 255, cv.THRESH_BINARY | cv.THRESH_TRIANGLE)
    print("threshold value %s" % ret)
    cv.imshow("Binary IMage", binary)


image = cv.resize(cv.imread("01.jpg"), (300,400))
cv.imshow("image", image)
threshold_demo(image)
cv.waitKey(0)
cv.destroyAllWindows()

threshold value 189.0

1.3.3 自动与手动

import cv2 as cv
import numpy


def threshold_demo(image):
    gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
    ret, binary = cv.threshold(gray, 150, 255, cv.THRESH_BINARY)
    print("threshold value %s" % ret)
    cv.imshow("Binary IMage", binary)


image = cv.resize(cv.imread("01.jpg"), (300,400))
cv.imshow("image", image)
threshold_demo(image)
cv.waitKey(0)
cv.destroyAllWindows()

threshold value 150.0

1.3.4 局部阈值

• 使用ADAPTIVE_THRESH_MEAN_C

import cv2 as cv
import numpy

def local_threshold_demo(image):
    gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
    binary = cv.adaptiveThreshold(gray, 255, cv.ADAPTIVE_THRESH_MEAN_C, cv.THRESH_BINARY, 25, 10)
    cv.imshow("Binary IMage", binary)

image = cv.resize(cv.imread("01.jpg"), (300,400))
cv.imshow("image", image)
local_threshold_demo(image)
cv.waitKey(0)
cv.destroyAllWindows()

• 使用ADAPTIVE_THRESH_GAUSSIAN_C

import cv2 as cv
import numpy

def local_threshold_demo(image):
    gray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
    binary = cv.adaptiveThreshold(gray, 255, cv.ADAPTIVE_THRESH_GAUSSIAN_C, cv.THRESH_BINARY, 25, 10)
    cv.imshow("Binary IMage", binary)

image = cv.resize(cv.imread("01.jpg"), (300,400))
cv.imshow("image", image)
local_threshold_demo(image)
cv.waitKey(0)
cv.destroyAllWindows()

2. 图像金字塔

图像金字塔原理

• 图像金字塔是图像多尺度表达的一种，是一种以多分辨率来解释图像的有效但概念简单的结构。
• 一幅图像的金字塔是一系列以金字塔形状排列的分辨率逐步降低，且来源于同一张原始图的图像集合。其通过梯次向下采样获得，直到达到某个终止条件才停止采样。
• 我们将一层一层的图像比喻成金字塔，层级越高，则图像越小，分辨率越低。

高斯金字塔和拉普拉斯金字塔

• Reduce：先对图像进行高斯模糊，再取出所有行和列中的偶数行和偶数列进行降采样

• Expand：先对图像进行扩大，再对扩大的图像进行卷积

import cv2 as cv


def pyramid_demo(image):
   level = 3
   temp = image.copy()
   pyramid_images = []
   for i in range(level):
       dst = cv.pyrDown(temp)
       pyramid_images.append(dst)
       cv.imshow("pyramid_down_" + str(i), dst)
       temp = dst.copy()
   return pyramid_images


def ladpalian_demo(image):
   pyramid_images = pyramid_demo(image)
   level = len(pyramid_images)
   for i in range(level - 1, -1, -1):
       if (i - 1) < 0:
           expand = cv.pyrUp(pyramid_images[i], dstsize=image.shape[:2])
           lpls = cv.subtract(image, expand)
           cv.imshow("laplalian_down_" + str(i), lpls)
       else:
           expand = cv.pyrUp(pyramid_images[i], dstsize=pyramid_images[i - 1].shape[:2])
           lpls = cv.subtract(pyramid_images[i - 1], expand)
           cv.imshow("laplalian_down_" + str(i), lpls)

image = cv.imread("01.jpg")
pyramid_demo(image)
cv.waitKey(0)
cv.destroyAllWindows()

• pyramid_demo

• ladpalian_demo