def GaussianNoise(self, filename1, filename2):
bmp = ReadBmp(filename1)
# 生成均值为means，方差为sigma，噪声比为percetage的高斯噪声图像
means = 8
sigma = 10
percetage = 1
for pixel in bmp.data:
# 图像灰度化
gray = 0.299 * pixel[0] + 0.578 * pixel[1] + 0.114 * pixel[2]
p = random.random()
# 增加高斯噪声
if p <= percetage:
gray = gray + random.gauss(means, sigma)
if gray < 0:
gray = 0
elif gray > 255:
gray = 255
pixel[0] = round(gray)
pixel[1] = round(gray)
pixel[2] = round(gray)
# 将新图像保存到文件
bmp.creataBmp(filename2)
return
def MedianFilter(self, filename1, filename2):
# 中值滤波，这里对边界填充采用不做边界处理的方式
bmp = ReadBmp(filename1)
new_data = []
for j in range(bmp.biHeight):
for i in range(bmp.biWidth):
if i == 0 or j == 0 or i == bmp.biWidth-1 or j == bmp.biHeight-1:
new_data.append(bmp.data[j * bmp.biWidth + i][0])
continue
list = [bmp.data[j * bmp.biWidth + i + 1][0], bmp.data[j * bmp.biWidth + i - 1][0] \
, bmp.data[(j + 1) * bmp.biWidth + i][0], bmp.data[(j - 1) * bmp.biWidth + i][0] \
, bmp.data[(j + 1) * bmp.biWidth + i + 1][0], bmp.data[(j + 1) * bmp.biWidth + i - 1][0] \
, bmp.data[(j - 1) * bmp.biWidth + i + 1][0], bmp.data[(j - 1) * bmp.biWidth + i - 1][0]]
list.sort()
median = round((list[3] + list[4])/2)
new_data.append(median)
for i in range(len(new_data)):
bmp.data[i][0] = new_data[i]
bmp.data[i][1] = new_data[i]
bmp.data[i][2] = new_data[i]
# 将新图像保存到文件
bmp.creataBmp(filename2)
return
def Sobel(self, filename1, filename2):
bmp = ReadBmp(filename1)
# 阈值，根据实验结果大致调整的
Gmax = 140
for i in range(bmp.biHeight * bmp.biWidth):
# 图像灰度化
gray = round(0.299 * bmp.data[i][0] + 0.578 * bmp.data[i][1] +
0.114 * bmp.data[i][2])
bmp.data[i][0] = gray
bmp.data[i][1] = gray
bmp.data[i][2] = gray
new_data = []
for j in range(bmp.biHeight):
for i in range(bmp.biWidth):
if i == 0 or j == 0 or i == bmp.biWidth - 1 or j == bmp.biHeight - 1:
new_data.append(0)
continue
Gx = (-1)*bmp.data[(j + 1) * bmp.biWidth + i - 1][0] + bmp.data[(j + 1) * bmp.biWidth + i + 1][0] \
+ (-1)*bmp.data[(j - 1) * bmp.biWidth + i - 1][0] + bmp.data[(j - 1) * bmp.biWidth + i + 1][0] \
+ (-2)*bmp.data[j * bmp.biWidth + i - 1][0] + 2*bmp.data[j * bmp.biWidth + i + 1][0]
Gy = (-1)*bmp.data[(j - 1) * bmp.biWidth + i - 1][0] + bmp.data[(j + 1) * bmp.biWidth + i - 1][0] \
+ (-1)*bmp.data[(j - 1) * bmp.biWidth + i + 1][0] + bmp.data[(j + 1) * bmp.biWidth + i + 1][0] \
+ (-2)*bmp.data[(j - 1) * bmp.biWidth + i][0] + 2*bmp.data[(j + 1) * bmp.biWidth + i][0]
G = math.sqrt(Gx**2 + Gy**2)
if G >= Gmax:
new_data.append(255)
else:
new_data.append(0)
for i in range(len(new_data)):
bmp.data[i][0] = new_data[i]
bmp.data[i][1] = new_data[i]
bmp.data[i][2] = new_data[i]
# 将新图像保存到文件
bmp.creataBmp(filename2)
return
def SaltAndPepperNoise(self, filename1, filename2):
bmp = ReadBmp(filename1)
# 生成椒盐噪声图像
percetage = 0.05
for pixel in bmp.data:
# 图像灰度化
gray = 0.299 * pixel[0] + 0.578 * pixel[1] + 0.114 * pixel[2]
p = random.random()
# 增加椒盐噪声
if p <= percetage:
if p <= percetage * 0.5:
gray = 0
else:
gray = 255
pixel[0] = round(gray)
pixel[1] = round(gray)
pixel[2] = round(gray)
# 将新图像保存到文件
bmp.creataBmp(filename2)
return
def MeanFilter(self, filename1, filename2):
# 均值滤波， 这里对边界填充采用不做边界处理的方式
bmp = ReadBmp(filename1)
new_data = []
for j in range(bmp.biHeight):
for i in range(bmp.biWidth):
if i == 0 or j == 0 or i == bmp.biWidth-1 or j == bmp.biHeight-1:
new_data.append(bmp.data[j * bmp.biWidth + i][0])
continue
sum = bmp.data[j * bmp.biWidth + i + 1][0] + bmp.data[j * bmp.biWidth + i - 1][0] \
+ bmp.data[(j + 1) * bmp.biWidth + i][0] + bmp.data[(j - 1) * bmp.biWidth + i][0] \
+ bmp.data[(j + 1) * bmp.biWidth + i + 1][0] + bmp.data[(j + 1) * bmp.biWidth + i - 1][0] \
+ bmp.data[(j - 1) * bmp.biWidth + i + 1][0] + bmp.data[(j - 1) * bmp.biWidth + i - 1][0]
new_data.append(round(sum/8))
for i in range(len(new_data)):
bmp.data[i][0] = new_data[i]
bmp.data[i][1] = new_data[i]
bmp.data[i][2] = new_data[i]
# 将新图像保存到文件
bmp.creataBmp(filename2)
return
from ReadBmp import ReadBmp
import matplotlib.pyplot as plt
import numpy as np
filename1 = "1.bmp"
bmp = ReadBmp(filename1)
bmp.gray()
# 统计各像素点数
h = np.array([0 for i in range(256)])
h1 = []
for pixel in bmp.data:
h[pixel[0]] = h[pixel[0]] + 1
h1.append(pixel[0])
# 画出原先的直方图
plt.subplot(1, 2, 1)
plt.hist(h1, bins=256)
# 归一化
hs = h / len(bmp.data)
# 计算累计分布
hp = np.array([0.0 for i in range(256)])
for i in range(256):
hp[i] = np.round(np.sum(hs[0:i + 1]) * 255)
T = hp.astype('uint8')
# 创建新图像，并统计新图像的各个像素点的个数
hn = np.array([0 for i in range(256)])
h2 = []
for pixel in bmp.data:
s = T[pixel[0]]