Python图像处理之图像量化处理详解

# -*- coding: utf-8 -*-
# By：Eastmount
import cv2  
import numpy as np  
import matplotlib.pyplot as plt

#读取原始图像
img = cv2.imread('lena-hd.png')

#获取图像高度和宽度
height = img.shape[0]
width = img.shape[1]

#创建一幅图像
new_img = np.zeros((height, width, 3), np.uint8)

#图像量化操作 量化等级为2
for i in range(height):
    for j in range(width):
        for k in range(3): #对应BGR三分量
            if img[i, j][k] < 128:
                gray = 0
            else:
                gray = 128
            new_img[i, j][k] = np.uint8(gray)
        
#显示图像
cv2.imshow("src", img)
cv2.imshow("", new_img)

#等待显示
cv2.waitKey(0)
cv2.destroyAllWindows()

# -*- coding: utf-8 -*-
# By：Eastmount
import cv2  
import numpy as np  
import matplotlib.pyplot as plt

#读取原始图像
img = cv2.imread('lena-hd.png')

#获取图像高度和宽度
height = img.shape[0]
width = img.shape[1]

#创建一幅图像
new_img1 = np.zeros((height, width, 3), np.uint8)
new_img2 = np.zeros((height, width, 3), np.uint8)
new_img3 = np.zeros((height, width, 3), np.uint8)

#图像量化等级为2的量化处理
for i in range(height):
    for j in range(width):
        for k in range(3): #对应BGR三分量
            if img[i, j][k] < 128:
                gray = 0
            else:
                gray = 128
            new_img1[i, j][k] = np.uint8(gray)

#图像量化等级为4的量化处理
for i in range(height):
    for j in range(width):
        for k in range(3): #对应BGR三分量
            if img[i, j][k] < 64:
                gray = 0
            elif img[i, j][k] < 128:
                gray = 64
            elif img[i, j][k] < 192:
                gray = 128
            else:
                gray = 192
            new_img2[i, j][k] = np.uint8(gray)

#图像量化等级为8的量化处理
for i in range(height):
    for j in range(width):
        for k in range(3): #对应BGR三分量
            if img[i, j][k] < 32:
                gray = 0
            elif img[i, j][k] < 64:
                gray = 32
            elif img[i, j][k] < 96:
                gray = 64
            elif img[i, j][k] < 128:
                gray = 96
            elif img[i, j][k] < 160:
                gray = 128
            elif img[i, j][k] < 192:
                gray = 160
            elif img[i, j][k] < 224:
                gray = 192
            else:
                gray = 224
            new_img3[i, j][k] = np.uint8(gray)

#用来正常显示中文标签
plt.rcParams['font.sans-serif']=['SimHei']

#显示图像
titles = ['(a) 原始图像', '(b) 量化-L2', '(c) 量化-L4', '(d) 量化-L8']  
images = [img, new_img1, new_img2, new_img3]  
for i in range(4):  
   plt.subplot(2,2,i+1), plt.imshow(images[i], 'gray'), 
   plt.title(titles[i])  
   plt.xticks([]),plt.yticks([])  
plt.show()

# coding: utf-8
# By：Eastmount
import cv2
import numpy as np
import matplotlib.pyplot as plt

#读取原始图像
img = cv2.imread('luo.png') 

#图像二维像素转换为一维
data = img.reshape((-1,3))
data = np.float32(data)

#定义中心 (type,max_iter,epsilon)
criteria = (cv2.TERM_CRITERIA_EPS +
            cv2.TERM_CRITERIA_MAX_ITER, 10, 1.0)

#设置标签
flags = cv2.KMEANS_RANDOM_CENTERS

#K-Means聚类 聚集成4类
compactness, labels, centers = cv2.kmeans(data, 8, None, criteria, 10, flags)


#图像转换回uint8二维类型
centers = np.uint8(centers)
res = centers[labels.flatten()]
dst = res.reshape((img.shape))

#图像转换为RGB显示
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
dst = cv2.cvtColor(dst, cv2.COLOR_BGR2RGB)


#用来正常显示中文标签
plt.rcParams['font.sans-serif']=['SimHei']

#显示图像
titles = ['原始图像', '聚类量化 K=8']  
images = [img, dst]  
for i in range(2):  
   plt.subplot(1,2,i+1), plt.imshow(images[i], 'gray'), 
   plt.title(titles[i])  
   plt.xticks([]),plt.yticks([])  
plt.show()