直方图均衡化
一、定义及作用
定义:直方图均衡化是将原图像通过某种变换,得到一幅灰度直方图为均匀分布的新图像的方法。
作用:直方图均衡化方法的基本思想是对在图像中像素个数多的灰度级进行展宽,而对像素个数少的灰度级进行缩减。从而达到清晰图像的目的。
二、理论原理
1、映射函数的条件
为了将原图像的亮度范围进行扩展,需要一个映射函数,将原图像的像素值均衡映射到新直 方图中,根据直方图的定义,这个映射函数有两个条件:
1. 为了不打乱原有的顺序,映射后亮、暗的大小关系不能改变,
2. 映射后必须在原有的范围内,比如(0-255)
2、映射函数的选取
1. 对于输入图像的任意一个像素p, p∈[0,255], 总能在输出图像里有对应的像素q, q∈[0,255] 使得下面等式成 立 (输入和输出的像素总量相等) :
其中、hist(k)即灰度级为k的像素个数2. 其中,输出图像每个灰度级的个数(我们所希望的映射输出):3.代入累加直方图公式,反求出对应关系:
计算实例
python实现
import cv2
import matplotlib.pyplot as plt
def histogram(img_gray):
'''
求灰度图像的直方图
:param img_gray: 灰度图像
:return:
'''
histogram = [0 for x in range(256)]
h,w = img_gray.shape
for i in range(h):
for j in range(w):
histogram[img_gray[i][j]] = histogram[img_gray[i][j]]+1
return histogram
def histogram_equalization(img_gray):
'''
灰度图直方图均衡化
:param img_gray: 灰度图像
:return: 灰度图直方图均衡化后的图像
'''
histogram_data = histogram(img_gray)
Pi = [0 for x in range(256)]
SumPi = []
sum = 0
h, w = img_gray.shape
total_pixels = h*w
for i in range(len(histogram_data)):
Pi[i] = histogram_data[i] / total_pixels
sum = sum + Pi[i]
SumPi.append(sum)
equalization = {}
for i in range(len(histogram_data)):
equalization[i] = round(SumPi[i] * 256-1 )
# return equalization
for i in range(h):
for j in range(w):
img_gray[i][j] = equalization[img_gray[i][j]]
return img_gray
# 设置matplotlib正常显示中文和负号
plt.rcParams['font.sans-serif']=['SimHei'] # 用黑体显示中文
plt.rcParams['axes.unicode_minus']=False # 正常显示负号
img = cv2.imread('lenna.png',1)
img_gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
plt.subplot(221)
plt.imshow(img_gray,cmap = "gray")
plt.title("灰度图")
plt.subplot(222)
hist_x = [x for x in range(256)]
plt.xticks(range(0,255,25))
hist = histogram(img_gray)
plt.bar(hist_x,hist,width=1)
plt.title("灰度直方图")
plt.subplot(223)
img_equal = histogram_equalization(img_gray)
plt.imshow(img_equal,cmap = "gray")
plt.title("均衡化")
plt.subplot(224)
hist_equal = histogram(img_equal)
plt.bar(hist_x,hist_equal,width=1)
plt.title("均衡化后的直方图")
plt.show()
def RGB_histogram(img):
'''
求彩色图像的直方图
:param img: BGR图像或者RGB图像
:return:
'''
Bhistogram = histogram(img[:,:,0])
Ghistogram = histogram(img[:,:,1])
Rhistogram = histogram(img[:,:,2])
return Bhistogram,Ghistogram,Rhistogram
def RGB_histogram_equalization(img):
img[:, :, 0] = histogram_equalization(img[:,:,0])
img[:, :, 1] = histogram_equalization(img[:,:,1])
img[:, :, 2] = histogram_equalization(img[:,:,2])
return img
plt.subplot(221)
img1 = cv2.cvtColor(img,cv2.COLOR_BGR2RGB)
plt.imshow(img1)
plt.title("彩色图")
plt.subplot(222)
BGRhistogram = RGB_histogram(img)
plt.bar(hist_x,BGRhistogram[0],color="blue",width=1)
plt.bar(hist_x,BGRhistogram[1],color="green",width=1)
plt.bar(hist_x,BGRhistogram[2],color="red",width=1)
plt.title("彩色图的直方图")
plt.subplot(223)
RGB_img_equal = RGB_histogram_equalization(img)
RGB_img_equal1 = cv2.cvtColor(RGB_img_equal,cv2.COLOR_BGR2RGB)
plt.imshow(RGB_img_equal1)
plt.title("彩色图均衡化")
plt.subplot(224)
RGB_hist_equal = RGB_histogram(RGB_img_equal)
plt.bar(hist_x,RGB_hist_equal[0],color="blue",width=1)
plt.bar(hist_x,RGB_hist_equal[1],color="green",width=1)
plt.bar(hist_x,RGB_hist_equal[2],color="red",width=1)
plt.title("彩色图均衡化后的直方图")
plt.show()
cv2.imshow("RGB_img_equal",RGB_img_equal)
cv2.waitKey(0)