• evm 代码分析

    处理时：

    • 为了放大运动，我们推荐我们新的基于相位的管道（SIGGRAPH 2013）。
    • 要放大颜色，请使用线性管道（SIGGRAPH 2012）。
    • 选择您想要放大的正确时间尺度。 例如，成年人的心跳往往每秒发生一次，对应于 1Hz，您可以将内容放大 0.5Hz 到 3Hz 以确保安全。 间隔越窄，放大越集中，噪声被放大的越少，但有遗漏物理现象的风险。
    • 在指定时间通带时不要忘记考虑视频帧速率！ 有关示例，请参见我们的代码。

    先介绍2012相位欧拉代码

    reproduceResultsSiggraph12

    % Generates all the results for the SIGGRAPH paper at:
% http://people.csail.mit.edu/mrub/vidmag
%
% Copyright (c) 2011-2012 Massachusetts Institute of Technology, 
% Quanta Research Cambridge, Inc.
%
% Authors: Hao-yu Wu, Michael Rubinstein, Eugene Shih
% License: Please refer to the LICENCE file
% Date: June 2012
%

clear;
dataDir = './data';
resultsDir = 'ResultsSIGGRAPH2012';

mkdir(resultsDir);

%% baby
% inFile = fullfile(dataDir,'baby.mp4'); % 路径拼接
% fprintf('Processing %s\n', inFile);
% amplify_spatial_lpyr_temporal_iir(inFile, resultsDir, 10, 16, 0.4, 0.05, 0.1);
% amplify_spatial_lpyr_temporal_iir 函数是一种放大算法，即空间滤波采用 拉普拉斯金字塔，时域滤波采用 IIR滤波器。

% Alternative processing using butterworth filter  使用巴特沃斯滤波器的替代处理
% amplify_spatial_lpyr_temporal_butter(inFile, resultsDir, 30, 16, 0.4, 3, 30, 0.1);
% 空间滤波采用 lpyr 拉普拉斯金字塔，时域滤波采用 butter 巴特沃斯滤波器

%% baby2
% inFile = fullfile(dataDir,'baby2.mp4');
% fprintf('Processing %s\n', inFile);
% amplify_spatial_Gdown_temporal_ideal(inFile,resultsDir,150,6, 140/60,160/60,30, 1);
% 空间滤波采用 Gaussian金字塔  时域滤波采用 Ideal bandpass

%% camera
% inFile = fullfile(dataDir,'camera.mp4');
% fprintf('Processing %s\n', inFile);
% amplify_spatial_lpyr_temporal_butter(inFile, resultsDir, 150, 20, 45, 100, 300, 0);
% 空间滤波采用 lpyr 拉普拉斯金字塔，时域滤波采用 butter 巴特沃斯滤波器


%% subway
% inFile = fullfile(dataDir,'subway.mp4');
% fprintf('Processing %s\n', inFile);
% amplify_spatial_lpyr_temporal_butter(inFile, resultsDir, 60, 90, 3.6, 6.2, 30, 0.3);
% 空间滤波采用 lpyr 拉普拉斯金字塔，时域滤波采用 butter 巴特沃斯滤波器

%% wrist
%% No mask is used here to generate the output video.
% inFile = fullfile(dataDir,'wrist.mp4');
% fprintf('Processing %s\n', inFile);
% amplify_spatial_lpyr_temporal_iir(inFile, resultsDir, 10, 16, 0.4, 0.05, 0.1);
% 空间滤波采用 lpyr 拉普拉斯金字塔，时域滤波采用 IIR滤波器

% Alternative processing using butterworth filter
% amplify_spatial_lpyr_temporal_butter(inFile, resultsDir, 30, 16, 0.4, 3, 30, 0.1);
% 空间滤波采用 lpyr 拉普拉斯金字塔，时域滤波采用 butter 巴特沃斯滤波器


%% shadow
% inFile = fullfile(dataDir,'shadow.mp4');
% fprintf('Processing %s\n', inFile);
% amplify_spatial_lpyr_temporal_butter(inFile, resultsDir, 5, 48, 0.5, 10, 30, 0);
% 空间滤波采用 lpyr 拉普拉斯金字塔，时域滤波采用 butter 巴特沃斯滤波器

%% guitar
% inFile = fullfile(dataDir,'guitar.mp4');
% fprintf('Processing %s\n', inFile);
% % amplify E
% amplify_spatial_lpyr_temporal_ideal(inFile, resultsDir, 50, 10, 72, 92, 600, 0);
% 空间滤波采用 lpyr 拉普拉斯金字塔，时域滤波采用 Ideal bandpass
% % amplify A
% amplify_spatial_lpyr_temporal_ideal(inFile, resultsDir, 100, 10, 100, 120, 600, 0);
% 空间滤波采用 lpyr 拉普拉斯金字塔，时域滤波采用 Ideal bandpass


%% face
% inFile = fullfile(dataDir,'myface.mp4');
% fprintf('Processing %s\n', inFile);
% amplify_spatial_Gdown_temporal_ideal(inFile,resultsDir,50,4, ...
%                      50/60,60/60,30, 1);
% 空间滤波采用 Gaussian金字塔  时域滤波采用 Ideal bandpass


%% face2
inFile = fullfile(dataDir,'face2.mp4');
fprintf('Processing %s\n', inFile);
% 
% % Motion 动作放大
% amplify_spatial_lpyr_temporal_butter(inFile,resultsDir,20,80, ...
%                                      0.5,10,30, 0);
% 空间滤波采用 lpyr 拉普拉斯金字塔，时域滤波采用 butter 巴特沃斯滤波器

% % Color 颜色放大
amplify_spatial_Gdown_temporal_ideal(inFile,resultsDir,50,6, ...
                                   50/60,60/60,30, 1);
% amplify_spatial_Gdown_temporal_ideal(inFile,resultsDir,120,6, ...
%                                    0.8,1,30, 1);
% 空间滤波采用 Gaussian金字塔  时域滤波采用 Ideal bandpass

    • 1
    • 2
    • 3
    • 4
    • 5
    • 6
    • 7
    • 8
    • 9
    • 10
    • 11
    • 12
    • 13
    • 14
    • 15
    • 16
    • 17
    • 18
    • 19
    • 20
    • 21
    • 22
    • 23
    • 24
    • 25
    • 26
    • 27
    • 28
    • 29
    • 30
    • 31
    • 32
    • 33
    • 34
    • 35
    • 36
    • 37
    • 38
    • 39
    • 40
    • 41
    • 42
    • 43
    • 44
    • 45
    • 46
    • 47
    • 48
    • 49
    • 50
    • 51
    • 52
    • 53
    • 54
    • 55
    • 56
    • 57
    • 58
    • 59
    • 60
    • 61
    • 62
    • 63
    • 64
    • 65
    • 66
    • 67
    • 68
    • 69
    • 70
    • 71
    • 72
    • 73
    • 74
    • 75
    • 76
    • 77
    • 78
    • 79
    • 80
    • 81
    • 82
    • 83
    • 84
    • 85
    • 86
    • 87
    • 88
    • 89
    • 90
    • 91
    • 92
    • 93
    • 94
    • 95
    • 96
    • 97
    • 98

    线性颜色和运动放大（SIGGRAPH 2012）

    该代码包括以下空间和时间过滤器的组合
    在这里插入图片描述
    拉普拉斯金字塔理想带通
    拉普拉斯金字塔巴特沃斯带通
    拉普拉斯金字塔二阶 IIR 带通
    高斯金字塔理想带通
    在这里插入图片描述

    以上是2012线性欧拉相关信息

    下面介绍2013相位欧拉代码

    reproduceResultsSiggraph13

    clear;
dataDir = './data';
resultsDir = 'ResultsSIGGRAPH2013/';
mkdir(resultsDir);

defaultPyrType = 'halfOctave'; % Half octave pyramid is default as discussed in paper 
% 如论文中所述，默认为半倍频程金字塔
scaleAndClipLargeVideos = true; % With this enabled, approximately 4GB of memory is used
% 启用此功能后，将使用大约 4GB 的内存

% Uncomment to use octave bandwidth pyramid: speeds up processing,
% but will produce slightly different results
%取消注释以使用倍频程带宽金字塔：加速处理，但会产生略有不同的结果
%defaultPyrType = 'octave'; 

% Uncomment to process full video sequences (uses about 16GB of memory)
% 取消注释以处理完整的视频序列（使用大约 16GB 的内存）
%scaleAndClipLargeVideos = false;

%% Car Engine
% inFile = fullfile(dataDir, 'car_engine.avi');
% samplingRate = 400; % Hz
% loCutoff = 15;    % Hz
% hiCutoff = 25;    % Hz
% alpha = 15;    
% sigma = 3;         % Pixels 像素
% pyrType = 'octave';
% if (scaleAndClipLargeVideos)
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'pyrType', pyrType,'scaleVideo', 0.5);
% else
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'pyrType', pyrType,'scaleVideo', 1);
% end
% % The sequence is very large. To save on CPU time, we set
% % pyramid type to 'octave'. If you have the resources or time, feel free to change it
% % to 'halfOctave'
% 序列非常大。 为了节省 CPU 时间，我们设置金字塔类型为“八度”。 如果您有资源或时间，请随时将其更改为“halfOctave”

%% Crane
% inFile = fullfile(dataDir, 'crane.avi');
% samplingRate = 24; % Hz
% loCutoff = 0.2;    % Hz
% hiCutoff = 0.25;    % Hz
% alpha = 100;    
% sigma = 5;         % Pixels
% temporalFilter = @FIRWindowBP; 
% pyrType = defaultPyrType;
% if (scaleAndClipLargeVideos)
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'pyrType', pyrType,'temporalFilter', temporalFilter,'scaleVideo', 2/3);
% else
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'pyrType', pyrType,'temporalFilter', temporalFilter, 'scaleVideo', 1);
% end  


% Crane Crop
% inFile = fullfile(dataDir, 'crane_crop.avi');
% samplingRate = 24; % Hz
% loCutoff = 0.2;    % Hz
% hiCutoff = 0.25;    % Hz
% alpha = 75;    
% sigma = 5;         % Pixels
% temporalFilter = @FIRWindowBP; 

% Comparison of cropped crane  起重机比较
% pyrType = 'octave';
% phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'pyrType', pyrType,'temporalFilter', temporalFilter);
% 
% pyrType = 'halfOctave';
% phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'pyrType', pyrType,'temporalFilter', temporalFilter);
% 
% pyrType = 'quarterOctave';
% phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'pyrType', pyrType,'temporalFilter', temporalFilter);

%% Eye
% inFile = fullfile(dataDir,  'eye.avi');
% samplingRate = 500; % Hz
% loCutoff = 30;    % Hz
% hiCutoff = 50;    % Hz
% alpha = 75;    
% sigma = 4;         % Pixels
% pyrType = 'octave';
% 
% if (scaleAndClipLargeVideos)
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'pyrType', pyrType, 'scaleVideo', 0.4);
% else
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'pyrType', pyrType, 'scaleVideo', 1);
% end

% The sequence is very large. To save on memory and CPU time, we set
% pyramid type to 'octave'. If you have the resources, feel free to change it
% to 'halfOctave'

%% Trees
% inFile = fullfile(dataDir, 'trees.avi');
% samplingRate = 60; % Hz
% alpha = 25;
% sigma = 2;         % Pixels 
% attenuateOtherFrequencies = true;
% pyrType = defaultPyrType;

% Low frequencies
% loCutoff = 0.5;    % Hz
% hiCutoff = 1;    % Hz
% phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'attenuateOtherFreq', attenuateOtherFrequencies,'pyrType', pyrType);

% Middle frequencies
% loCutoff = 1.5;
% hiCutoff = 2;
% phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'attenuateOtherFreq', attenuateOtherFrequencies,'pyrType', pyrType);

%% Throat
% inFile = fullfile(dataDir, 'throat.avi');
% samplingRate = 1900; % Hz
% loCutoff = 90;    % Hz
% hiCutoff = 110;    % Hz
% alpha = 100;    
% sigma = 3;         % Pixels
% pyrType = 'octave';
% 
% if (scaleAndClipLargeVideos)
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'pyrType', pyrType, 'scaleVideo', 2/3);
% else
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'pyrType', pyrType, 'scaleVideo', 1);
% end

%% Woman
% inFile = fullfile(dataDir, 'woman.avi');
% samplingRate = 60; % Hz
% sigma = 3;
% alpha = 15;
% attenuateOtherFrequencies = true;
% pyrType = defaultPyrType;
% 
% % Low frequencies
% loCutoff = 0.35;
% hiCutoff = 0.71;
% if (scaleAndClipLargeVideos)
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'attenuateOtherFreq', attenuateOtherFrequencies,'pyrType', pyrType, 'scaleVideo', 0.9);
% else
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'attenuateOtherFreq', attenuateOtherFrequencies,'pyrType', pyrType, 'scaleVideo', 1);
% end

% Middle frequencies 中频
% loCutoff = 1;
% hiCutoff = 1.9;
% phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma,'attenuateOtherFreq', attenuateOtherFrequencies,'pyrType', pyrType);


%% Jumping Boy
% inFile = fullfile(dataDir, 'stomp.avi');
% phaseAmplifyLargeMotions; % 用于处理 stomp 视频而不处理大动作且不使用所有内存的脚本

%% Comparisons with Wu et al.

% % Baby
% inFile = fullfile(dataDir, 'baby41.avi');
% loCutoff = 0.04;
% hiCutoff = 0.4;
% temporalFilter = @differenceOfIIR;
% alpha = 20;
% sigma = 5;
% pyrType = 'quarterOctave';
% 
% if (scaleAndClipLargeVideos)
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, 1, resultsDir, 'sigma', sigma, 'pyrType', pyrType, 'temporalFilter', temporalFilter, 'scaleVideo', 0.8);
% else
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, 1, resultsDir, 'sigma', sigma, 'pyrType', pyrType, 'temporalFilter', temporalFilter, 'scaleVideo', 1);
% end

% Camera
% inFile = fullfile(dataDir, 'camera.mp4');
% samplingRate = 300;
% loCutoff = 36;
% hiCutoff = 62;
% temporalFilter = @differenceOfButterworths;
% alpha = 120;
% sigma = 5;
% pyrType = defaultPyrType;
% if (scaleAndClipLargeVideos)
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir, 'sigma', sigma, 'pyrType', pyrType, 'temporalFilter', temporalFilter, 'useFrames', [1 500]);
% else
%     phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir, 'sigma', sigma, 'pyrType', pyrType, 'temporalFilter', temporalFilter);
% end

% Guitar
% inFile = fullfile(dataDir, 'guitar.mp4');
% samplingRate = 600;
% loCutoff = 72;
% hiCutoff = 92;
% alpha = 25;
% sigma = 2;
% pyrType = defaultPyrType;
% phaseAmplify(inFile, alpha, loCutoff, hiCutoff, samplingRate, resultsDir,'sigma', sigma, 'pyrType', pyrType);

%% Motion Attenuation  运动衰减

% Moon
% inFile = fullfile(dataDir, 'moon.avi');
% outFile = fullfile(resultsDir, 'moon-motionAttenuated.avi');
% motionAttenuateMedianPhase(inFile, outFile);


% %Face
% inFile = fullfile(dataDir, 'myface.mp4');
% motAttFile = fullfile(resultsDir, 'face_motionattenuated.avi');
% motionAttenuateFixedPhase(inFile, motAttFile);
% amplify_spatial_Gdown_temporal_ideal(motAttFile,resultsDir,100,4,50/60,60/60,30, 1);


% Shuttle
% inFile = fullfile(dataDir, 'shuttle.avi');
% outFile = fullfile(resultsDir, 'shuttle-motionAttenuated.avi');
% loCutoff = 0.05;
% motionAttenuateLowpassPhase(inFile, outFile, loCutoff);


    • 1
    • 2
    • 3
    • 4
    • 5
    • 6
    • 7
    • 8
    • 9
    • 10
    • 11
    • 12
    • 13
    • 14
    • 15
    • 16
    • 17
    • 18
    • 19
    • 20
    • 21
    • 22
    • 23
    • 24
    • 25
    • 26
    • 27
    • 28
    • 29
    • 30
    • 31
    • 32
    • 33
    • 34
    • 35
    • 36
    • 37
    • 38
    • 39
    • 40
    • 41
    • 42
    • 43
    • 44
    • 45
    • 46
    • 47
    • 48
    • 49
    • 50
    • 51
    • 52
    • 53
    • 54
    • 55
    • 56
    • 57
    • 58
    • 59
    • 60
    • 61
    • 62
    • 63
    • 64
    • 65
    • 66
    • 67
    • 68
    • 69
    • 70
    • 71
    • 72
    • 73
    • 74
    • 75
    • 76
    • 77
    • 78
    • 79
    • 80
    • 81
    • 82
    • 83
    • 84
    • 85
    • 86
    • 87
    • 88
    • 89
    • 90
    • 91
    • 92
    • 93
    • 94
    • 95
    • 96
    • 97
    • 98
    • 99
    • 100
    • 101
    • 102
    • 103
    • 104
    • 105
    • 106
    • 107
    • 108
    • 109
    • 110
    • 111
    • 112
    • 113
    • 114
    • 115
    • 116
    • 117
    • 118
    • 119
    • 120
    • 121
    • 122
    • 123
    • 124
    • 125
    • 126
    • 127
    • 128
    • 129
    • 130
    • 131
    • 132
    • 133
    • 134
    • 135
    • 136
    • 137
    • 138
    • 139
    • 140
    • 141
    • 142
    • 143
    • 144
    • 145
    • 146
    • 147
    • 148
    • 149
    • 150
    • 151
    • 152
    • 153
    • 154
    • 155
    • 156
    • 157
    • 158
    • 159
    • 160
    • 161
    • 162
    • 163
    • 164
    • 165
    • 166
    • 167
    • 168
    • 169
    • 170
    • 171
    • 172
    • 173
    • 174
    • 175
    • 176
    • 177
    • 178
    • 179
    • 180
    • 181
    • 182
    • 183
    • 184
    • 185
    • 186
    • 187
    • 188
    • 189
    • 190
    • 191
    • 192
    • 193
    • 194
    • 195
    • 196
    • 197
    • 198
    • 199
    • 200
    • 201
    • 202
    • 203
    • 204
    • 205
    • 206
    • 207
    • 208
    • 209
    • 210
    • 211
    • 212
    • 213
    • 214

    基于相位的运动处理（SIGGRAPH 2013）
    该代码包括使用频域中的理想带通滤波器和任意时域 LTI 滤波器运动放大视频的功能。它还包括衰减运动的代码。
    在这里插入图片描述
    使用任意时间滤波器放大运动
    放大小动作，但保持大动作不变 (stomp.avi)
    通过将所有相位固定到参考框架来衰减运动 (face.avi)
    通过对相位应用中值滤波器来衰减运动 (moon.avi)
    在这里插入图片描述

  • 相关阅读:
    Web前端：Web 项目的前14个React图表库
    uniapp 模拟请求自身的JSON模拟数据
    iOS中UIColor，CGColor，CIColor详解
    深入理解Docker自定义网络：构建高效的容器网络环境
    秋招在线人才测评考什么内容？
    我和EarthSDK Cesium那点事（零）
    JAVA图书管理练习
    将你的桌面变成一个雨滴窗口：关于两个有趣的应用的整合
    Go/Golang语言学习实践[回顾]教程09--学习成绩统计的示例【上】
    使用JAVA CompletableFuture实现流水线化的并行处理，深度实践总结
  • 原文地址：https://blog.csdn.net/weixin_46118768/article/details/126332368