混合高斯背景建模matlab代码

clear all
% source = aviread('C:\Video\Source\traffic\san_fran_traffic_30sec_QVGA');
source = mmreader('SampleVideo.avi');
frameQYT=get(source,'NumberOfFrames');
% ----------------------- frame size variables -----------------------
fr = read(source,1); % 读取第一帧作为背景
fr_bw = rgb2gray(fr); % 将背景转换为灰度图像
fr_size = size(fr); %取帧大小
width = fr_size(2);
height = fr_size(1);
fg = zeros(height, width);
bg_bw = zeros(height, width);
% --------------------- mog variables -----------------------------------
C = 4; % 组成混合高斯的单高斯数目(一般3-5)
M = 0; % 组成背景的数目
D = 2.5; % 阈值（一般2.5个标准差）
alpha = 0.01; % learning rate 学习率决定更新速度(between 0 and 1) (from paper 0.01)
thresh = 0.75; % foreground threshold 前景阈值(0.25 or 0.75 in paper)
sd_init = 6; % initial standard deviation 初始化标准差(for new components) var = 36 in paper
w = zeros(height,width,C); % initialize weights array 初始化权值数组
mean = zeros(height,width,C); % pixel means 像素均值
sd = zeros(height,width,C); % pixel standard deviations 像素标准差
u_diff = zeros(height,width,C); % difference of each pixel from mean 与均值的差p = alpha/(1/C); % initial p variable 参数学习率(used to update mean and sd)
rank = zeros(1,C); % rank of components (w/sd)
% ------initialize component means and weights 初始化均值和权值----------
pixel_depth = 8; % 8-bit resolution 像素深度为8位
pixel_range = 2^pixel_depth -1; % pixel range 像素范围2的7次方0—255（# of possible values)
for i=1:height
for j=1:width
for k=1:C
mean(i,j,k) = rand*pixel_range; % means random (0-255之间的随机数)
w(i,j,k) = 1/C; % weights uniformly dist
sd(i,j,k) = sd_init; % initialize to sd_init
end
end
end
%----- process frames -处理帧--，这里去第八帧
n = 8;
fr = read(source,n); % read in frame 读取帧
fr_bw = rgb2gray(fr); % convert frame to grayscale 转换为灰度图像
% calculate difference of pixel values from mean 计算像素差值
for m=1:C
u_diff(:,:,m) = abs(double(fr_bw) - double(mean(:,:,m)));
end
% update gaussian components for each pixel 更新每个像素的背景模型
for i=1:height
for j=1:width
match = 0;
for k=1:C
if (abs(u_diff(i,j,k)) <= D*sd(i,j,k)) % pixel matches component像素匹配了模型
match = 1; % variable to signal component match 设置匹配记号
% update weights, mean, sd, p 更新权值，均值，标准差和参数学习率
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 % pixel doesn't match component 几个模型中都没有匹配的
w(i,j,k) = (1-alpha)*w(i,j,k); % weight slighly decreases 权值减小
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); %更新背景
if(bg_bw(i,j)>thresh)
k=k-1;
M=k;
end%???? 这里有问题，背景权值和大于阈值时，背景建模的数目M取k-1,
end
% if no components match, create new component 如果没有匹配的模型则创建新模型
if (match == 0)
[min_w, min_w_index] = min(w(i,j,:));
mean(i,j,min_w_index) = double(fr_bw(i,j));
sd(i,j,min_w_index) = sd_init;
end
rank = w(i,j,:)./sd(i,j,:); % calculate component rank 计算模型优先级
rank_ind = [1:1:C];
% calculate foreground 计算前景
while ((match == 0)&&(k>M))%????? 这里用于前景计算的高斯模型应该是C-M,所以这里k>M
if (abs(u_diff(i,j,rank_ind(k))) <= D*sd(i,j,rank_ind(k)))
fg(i,j) = 0; %black = 0
else
fg(i,j) = fr_bw(i,j);
end
k = k+1;
if(k==5)
k=k-1;
break
end
end
end
end
figure(1),subplot(3,1,1),imshow(fr) %显示输入图像
subplot(3,1,2),imshow(uint8(bg_bw)) %显示背景图像
subplot(3,1,3),imshow(uint8(fg)) %显示前景图像。