第一:下软件MATLAB，和插件SHINEtoolbox。安装路径最好不要有中文。

将插件复制到文件夹下见图片

注意：不用学习matlab的具体应用，会使用插件就行了，插件是已经做好的编程，仅需把插件的参数修改即可。

第二，加载插件，

进入matlab主页，点击设置路径。将路径设置为SHINE文件夹。设置路径后，记得点保存，才算加载成功。

在你在matlab命令窗口输入SHINE，当出现“SHINE options [1=default, 2=custom]: “

就说明调用成功(当然安装就成功了)。如果报错就不对。

第三，插件信息的修改。插件有很多函数，但是仅需要修改SHINE.m，需要你打开SHINE.m(双击即可)去手动设置。

具体修改的地方有

1.改为你处理的图片类型。如“png”

2.修改你要到处理图片的位置。

3.选用WholeIm=3时，要需要保证SHINE-INPUT和SHINE-TEMPLATE图片数量、格式、大小一致，否则报错。将处理的图片分别复制到两个文件夹下即可。

具体处理过程和参数设置

第四，处理图片

处理步骤：进入Matlab，加载“SHINEtoolbox”插件对图片的亮度、对比度和尺寸进行统一标准化批处理。

关键操作：输入SHINE→按Enter→输入1→按Enter。

图片处理成功会显示，图片会保存到SHINE-OUTPUT

附上图片库的处理参数设置：

【统一标准化批处理】

使用中国人面孔表情图片系统(罗跃嘉)的男正2.bmp图片作为模板(中国人面孔表情图片系统\bmp格式\正性\male\2.bmp)。

处理步骤：进入Matlab，加载“SHINEtoolbox”插件对图片的亮度、对比度和尺寸进行统一标准化批处理。

关键操作：输入SHINE→按Enter→输入1→按Enter。

>> SHINE

SHINE options [1=default, 2=custom]: 1

Number of images: 689

Option: Mean luminance matching separately for the foregrounds and backgrounds (background = all regions of lum 34)

Progress: lumMatch successful

RMSE: 2.327847e+01

SSIM: 8.104382e-01

默认模式(default)关键参数：图像背景分割(wholeIm = 3)+亮度匹配(mode = 1)，程序代码如下：

% If desired, the default values can be changed here:

it = 1; % number of iterations (default = 1)

wholeIm = 3; % 1 = whole image (default)

% 2 = figure-ground separated (input images as templates)

% 3 = figure-ground separated (based on templates: 中国人面孔表情图片系统\bmp格式\正性\male\2.bmp)

mode = 1; % 1 = lumMatch only (no SSIM optimization)(原版婴儿图库的处理)

% 2 = histMatch only(整体过亮)

% 3 = sfMatch only (比7稍微暗)

% 4 = specMatch only (有雾气)

% 5 = histMatch & sfMatch (比7局部过亮)

% 6 = histMatch & specMatch (有雾气+雪花)

% 7 = sfMatch & histMatch (整体较暗)

% 8 = specMatch & histMatch (default)(有雾气+雪花)

background = 300; % background lum of template, or 300=automatic (default)

% (automatically, the luminance that occurs most

% frequently in the image is used as background lum);

% basically, all regions of that lum are treated as

% background

rescaling = 1; % 0 = no rescaling

% 1 = rescale absolute values (default)

% 2 = rescale average values

optim = 1; % 0 = no SSIM optimization

% 1 = SSIM optimization (Avanaki, 2009; to change the

% number if iterations (default = 10) and adjust step

% size (default = 67), see histMatch.m)

【背景处理】

完成统一标准化批处理后，先将图片位深度调为24，然后在画图工具中对非面孔区域的背景进行黑色(RGB:0, 0, 0)填充，最后使用SHINEtoolbox对每张图片逐张标准化处理。

处理步骤：进入Matlab，加载“SHINEtoolbox”插件逐张对图片的亮度、对比度和尺寸进行标准化处理。

关键操作：输入SHINE→按Enter→输入1→按Enter。

>> SHINE

SHINE options [1=default, 2=custom]: 1

Number of images: 1

Option: Mean luminance matching on the whole images

Progress: lumMatch successful

RMSE: 0

SSIM: 1

默认模式(default)关键参数：图像背景分割(wholeIm = 1)+亮度匹配(mode = 1)，程序代码如下：

% If desired, the default values can be changed here:

it = 1; % number of iterations (default = 1)

wholeIm = 1; % 1 = whole image (default)

% 2 = figure-ground separated (input images as templates)

% 3 = figure-ground separated (based on templates)

mode = 1; % 1 = lumMatch only (no SSIM optimization)

% 2 = histMatch only

% 3 = sfMatch only

% 4 = specMatch only

% 5 = histMatch & sfMatch

% 6 = histMatch & specMatch

% 7 = sfMatch & histMatch

% 8 = specMatch & histMatch (default)

background = 300; % background lum of template, or 300=automatic (default)

% (automatically, the luminance that occurs most

% frequently in the image is used as background lum);

% basically, all regions of that lum are treated as

% background

rescaling = 1; % 0 = no rescaling

% 1 = rescale absolute values (default)

% 2 = rescale average values

optim = 1; % 0 = no SSIM optimization

% 1 = SSIM optimization (Avanaki, 2009; to change the

% number if iterations (default = 10) and adjust step

% size (default = 67), see histMatch.m)

【统一标准化批处理】

使用中国人面孔表情图片系统(罗跃嘉)的男正2.bmp图片作为模板(中国人面孔表情图片系统\bmp格式\正性\male\2.bmp)。

处理步骤：进入Matlab，加载“SHINEtoolbox”插件对图片的亮度、对比度和尺寸进行统一标准化批处理。

关键操作：输入SHINE→按Enter→输入1→按Enter。

>> SHINE

SHINE options [1=default, 2=custom]: 1

Number of images: 689

Option: Mean luminance matching separately for the foregrounds and backgrounds (background = all regions of lum 34)

Progress: lumMatch successful

RMSE: 2.327847e+01

SSIM: 8.104382e-01

默认模式(default)关键参数：图像背景分割(wholeIm = 3)+亮度匹配(mode = 1)，程序代码如下：

% If desired, the default values can be changed here:

it = 1; % number of iterations (default = 1)

wholeIm = 3; % 1 = whole image (default)

% 2 = figure-ground separated (input images as templates)

% 3 = figure-ground separated (based on templates: 中国人面孔表情图片系统\bmp格式\正性\male\2.bmp)

mode = 1; % 1 = lumMatch only (no SSIM optimization)(原版婴儿图库的处理)

% 2 = histMatch only(整体过亮)

% 3 = sfMatch only (比7稍微暗)

% 4 = specMatch only (有雾气)

% 5 = histMatch & sfMatch (比7局部过亮)

% 6 = histMatch & specMatch (有雾气+雪花)

% 7 = sfMatch & histMatch (整体较暗)

% 8 = specMatch & histMatch (default)(有雾气+雪花)

background = 300; % background lum of template, or 300=automatic (default)

% (automatically, the luminance that occurs most

% frequently in the image is used as background lum);

% basically, all regions of that lum are treated as

% background

rescaling = 1; % 0 = no rescaling

% 1 = rescale absolute values (default)

% 2 = rescale average values

optim = 1; % 0 = no SSIM optimization

% 1 = SSIM optimization (Avanaki, 2009; to change the

% number if iterations (default = 10) and adjust step

% size (default = 67), see histMatch.m)

【背景处理】

完成统一标准化批处理后，先将图片位深度调为24，然后在画图工具中对非面孔区域的背景进行黑色(RGB:0, 0, 0)填充，最后使用SHINEtoolbox对每张图片逐张标准化处理。

处理步骤：进入Matlab，加载“SHINEtoolbox”插件逐张对图片的亮度、对比度和尺寸进行标准化处理。

关键操作：输入SHINE→按Enter→输入1→按Enter。

>> SHINE

SHINE options [1=default, 2=custom]: 1

Number of images: 1

Option: Mean luminance matching on the whole images

Progress: lumMatch successful

RMSE: 0

SSIM: 1

默认模式(default)关键参数：图像背景分割(wholeIm = 1)+亮度匹配(mode = 1)，程序代码如下：

% If desired, the default values can be changed here:

it = 1; % number of iterations (default = 1)

wholeIm = 1; % 1 = whole image (default)

% 2 = figure-ground separated (input images as templates)

% 3 = figure-ground separated (based on templates)

mode = 1; % 1 = lumMatch only (no SSIM optimization)

% 2 = histMatch only

% 3 = sfMatch only

% 4 = specMatch only

% 5 = histMatch & sfMatch

% 6 = histMatch & specMatch

% 7 = sfMatch & histMatch

% 8 = specMatch & histMatch (default)

background = 300; % background lum of template, or 300=automatic (default)

% (automatically, the luminance that occurs most

% frequently in the image is used as background lum);

% basically, all regions of that lum are treated as

% background

rescaling = 1; % 0 = no rescaling

% 1 = rescale absolute values (default)

% 2 = rescale average values

optim = 1; % 0 = no SSIM optimization

% 1 = SSIM optimization (Avanaki, 2009; to change the

% number if iterations (default = 10) and adjust step

% size (default = 67), see histMatch.m)