%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Author: Ziheng H. Shen @Tsinghua Univ.
%HybridGaussModel @Digital Image Process Practice
%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
clc;
clear all;
% cntFrame = 23;
% obj = VideoReader('768x576.avi');
% numFrames = obj.NumberOfFrames;
% for k = 1 : cntFrame
% frame = read(obj,k);
% imwrite(frame,...
% strcat('C:\Users\Zi-Heng Shen\Documents\MATLAB\BackGroundModel\混合高斯背景建模\',...
% num2str(k),'.bmp'),'bmp');
% end
File=dir('D:\tracking\Walking\Walking\img\*.jpg');
%% 参数定义及初始化
I = imread('D:\tracking\Walking\Walking\img\0001.jpg'); %读入第一帧作为背景帧
fr_bw = I;
[height,width] = size(fr_bw); %求每帧图像大小
width = width/3; %排除颜色通道数
fg = zeros(height, width); %定义前景和背景矩阵
bg_bw = zeros(height, width);
C = 3; % 单高斯模型的个数(通常为3-5)
M = 3; % 代表背景的模型个数
D = 2.5; % 偏差阈值
alpha = 0.01; % 学习率
thresh = 0.25; % 前景阈值
sd_init = 15; % 初始化标准差
w = zeros(height,width,C); % 初始化权重矩阵
mean = zeros(height,width,C); % 像素均值
sd = zeros(height,width,C); % 像素标准差
u_diff = zeros(height,width,C); % 像素与某个高斯模型均值的绝对距离
p = alpha/(1/C); % 初始化p变量，用来更新均值和标准差
rank = zeros(1,C); % 各个高斯分布的优先级（w/sd)
pixel_depth = 8; % 每个像素8bit分辨率
pixel_range = 2^pixel_depth -1; % 像素值范围[0,255]
for i=1:height
for j=1:width
for k=1:C
mean(i,j,k) = rand*pixel_range; %初始化第k个高斯分布的均值
w(i,j,k) = 1/C; % 初始化第k个高斯分布的权重
sd(i,j,k) = sd_init; % 初始化第k个高斯分布的标准差
end
end
end
for n = 1:50
%frame=strcat(num2str(n),'.bmp');
%I1 = imread(frame); % 依次读入各帧图像
image_name = File(n).name;
image = imread(strcat('D:\tracking\Walking\Walking\img\',image_name));
fr_bw =image;
% 计算新像素与第m个高斯模型均值的绝对距离
for m=1:C
u_diff(:,:,m) = abs(double(fr_bw(:,:,m)) - double(mean(:,:,m)));
end
% 更新高斯模型的参数
for i=1:height
for j=1:width
match = 0; %匹配标记;
for k=1:C
if (abs(u_diff(i,j,k)) <= D*sd(i,j,k)) % 像素与第k个高斯模型匹配
match = 1; %将匹配标记置为1
% 更新权重、均值、标准差、p
w(i,j,k) = (1-alpha)*w(i,j,k) + alpha;
p = alpha/w(i,j,k);
mean(i,j,k) = (1-p)*mean(i,j,k) + p*double(fr_bw(i,j));
sd(i,j,k) = sqrt((1-p)*(sd(i,j,k)^2) + p*((double(fr_bw(i,j)) - mean(i,j,k)))^2);
else % 像素与第k个高斯模型不匹配
w(i,j,k) = (1-alpha)*w(i,j,k); %略微减少权重
end
end
bg_bw(i,j)=0;
for k=1:C
bg_bw(i,j) = bg_bw(i,j)+ mean(i,j,k)*w(i,j,k);
end
% 像素值与任一高斯模型都不匹配，则创建新的模型
if (match == 0)
[min_w, min_w_index] = min(w(i,j,:)); %寻找最小权重
mean(i,j,min_w_index) = double(fr_bw(i,j));%初始化均值为当前观测像素的均值
sd(i,j,min_w_index) = sd_init; %初始化标准差为6
end
rank = w(i,j,:)./sd(i,j,:); % 计算模型优先级
rank_ind = [1:1:C];%优先级索引
% 计算前景
fg(i,j) = 0;
while ((match == 0)&&(k<=M))
if (abs(u_diff(i,j,rank_ind(k))) <= D*sd(i,j,rank_ind(k)))% 像素与第k个高斯模型匹配
fg(i,j) = 0; %该像素为背景，置为黑色
else
fg(i,j) = 255; %否则为前景，置为白色
end
k = k+1;
end
end
end
figure(n)
subplot(1,3,1),imshow(fr_bw); %显示最后一帧图像
subplot(1,3,2),imshow(uint8(bg_bw)) %显示背景
disk = strel('disk',1);disk1 = strel('disk',4);
subplot(1,3,3),imshow(imdilate(imerode(uint8(fg),disk),disk1)); %显示前景
end
