直方圖均衡化的MATLAB實現
回顧----直方圖均衡化
- 基本原理
直方圖均衡化方法的基本思想是:
對在圖像中像素個數多的灰度級進行展寬,而對像素個數少的灰度級進行縮減,從而達到清晰圖像的目的
因爲灰度分佈可在直方圖中描述,所以該圖像增強方法是基於圖像的灰度直方圖。 - 直方圖均衡化的處理步驟
①求待處理圖像的直方圖h
②計算原圖的灰度分佈概率hs →
Nf--圖像f的總體像素個數 (m,n分別爲圖像的長和寬)
hs--每個灰度級的分佈概率,即每個像素在整個圖像中所佔的比例 (i=0,1,…,255)
③計算原圖灰度的累計分佈hp →
(i=0,1,…,255)
④計算原、新圖灰度值的影射關係
⑤原、新圖灰度直方圖比較
代碼實現
實現程序如下圖所示:
% 直方圖均衡化
% function [J] = imhisteq0(I)
function [J] = dip(I)
I = imread('img\person.jpg');
figure,imshow(I)
[m,n,l] = size(I);
if(l>1)
I = rgb2gray(I);
end
nbins = 256;
hist_0 = GetImHist0(I,nbins)'; %求直方圖
hist_1 = hist_0/(m*n); %求灰度分佈概率
hp_0 = cumsum(hist_1); %求原圖灰度累計分佈 MATLAB中cumsum可用於求累計和
hp_1 = round(hp_0*255);
hp_1(1) = 0; %第1個元素強制設爲0
I0 = double(I);
for i=1:m
for j=1:n
GrayScale = I0(i,j); %原圖灰度值
NewGrayScale = hp_1(GrayScale+1);
J(i,j) = NewGrayScale;
end
end
J = uint8(J);
figure,imshow(J)
end
% 直方圖
function counts = GetImHist0(Im,nbins)
% nbins箱子 區間的個數
[row,col,cChannel] = size(Im);
% cChannel顏色通道
counts = zeros(nbins,1);
minV = 0;
maxV = 255;
if(cChannel>1)
disp('Input error');
else
Im = double(Im);
Im2 = reshape(Im,row*col,1);
delta = (maxV-minV)/nbins;
splitVs = 0:nbins;
splitVs = splitVs*delta; %splitVs = linspace(minV,maxV,nbins+1)
i=1;
ind = find(Im2>=splitVs(i) & Im2<=splitVs(i+1));
counts(i) = length(ind);
for i = 2:nbins
ind = find(Im2>splitVs(i) & Im2<=splitVs(i+1));
counts(i) = length(ind);
end
end
end
代碼執行結果: