一个简单背景建模

看了看高斯背景建模

个人理解其实就是选出同一位置像素点出现概率最大的颜色值

这样的化，其实所建模的视频帧越多，则建模效果越好！

一个简单的代码：

clc;
clear;
d = 0:1:255;
result1 = zeros(768*576,256);
%result2 = zeros(768*576,256);
%result3 = zeros(768*576,256);
resultimage1 = zeros(576,768);
%resultimage2 = zeros(576,768);
%resultimage3 = zeros(576,768);
for k = 1:1:1
    for i=1:1:500
        i
        a = num2str(i);
        if i<10
            str = strcat('L:\PETS\S0_CC\Crowd_PETS09\S0\City_Center\Time_12-34\View_001\frame_000',a,'.jpg');
        else if i<100
            str = strcat('L:\PETS\S0_CC\Crowd_PETS09\S0\City_Center\Time_12-34\View_001\frame_00',a,'.jpg');
            else if i<1000
                    str = strcat('L:\PETS\S0_CC\Crowd_PETS09\S0\City_Center\Time_12-34\View_001\frame_0',a,'.jpg');
                end
            end
        end
        im = imread(str);
        tempimage1 = im(:,:,k);
        %tempimage2 = im(:,2);
        %tempimage3 = im(:,3);
        for j = 1:1:768*576
            n = find(tempimage1(j)==d);
            result1(j,n) = result1(j,n)+1;
        end
    %     for j = 1:1:768*576
    %         n = find(tempimage2(j)==d);
    %         result2(j,n) = result2(j,n)+1;
    %     end
    %     for j = 1:1:768*576
    %         n = find(tempimage3(j)==d);
    %         result3(j,n) = result3(j,n)+1;
    %     end
    end
    for j = 1:1:768*576
        t = max(result1(j,:));
        n = find(result1(j,:)==t);
        resultimage1(j) = n(1);
    end
    resultimage1 = resultimage1-1;
%     endimage(:,:,k) = resultimage1;
    endimage = resultimage1;
    %strk = int2str(k);
    %strend = strcat(k,'mat');
    %save 'strend' resultimage1;
end

图像帧采用的是pets2009中的s2的L1第一视角下的视频

大家可以看下结果：

效果凑活吧！