adaboost算法代码包

% 建立 特征编号 与特征名称 之间的对应关系 % featureNameMap % % 特征命名有如下例子（每一种方法提取的特征举一例）： % RGB-R-Mean：RGB颜色空间的R分量的均值 % HSI-H-Std： HSI颜色空间的H分量的标准差 % Gabor-50X22.5-Mean：以尺度50相位22.5作Gabor变换提取的均值特征 % 灰度共生矩阵-6X-6-Contrast：灰度共生矩阵方法在方向(6,-6)提取的对比度特征 % 灰度差分统计-10X10-Contrast：灰度差分统计方法在方向(10,10)提取的对比度特征 % % % 目前提取的颜色纹理特征值包括： % RGB颜色空间特征值 RGB颜色分量的均值,标准差,方差,相关系数 12个 % HSI颜色空间特征值 HSI颜色分量的均值,标准差,方差,相关系数 12个 % NTSC颜色空间特征值 YIQ颜色分量的均值,标准差,方差,相关系数 12个 % gabor变换提取的纹理特征 均值，标准差 % 特征数量 64=4(频率个数)X8(方向个数)X2(每组提取的纹理特征数量) % 灰度共生矩阵纹理特征值： % '角二阶矩','对比度','相关性','差分矩','逆差分矩（均匀性）','和平均值',... % '和方差','和熵','熵','差方差','差熵' % 特征数量 176= 16组位置算子 X 每组位置算子提取11个特征 % 灰度-梯度共生矩阵纹理特征，采用 roberts 算子提取梯度 % 提取15个特征，分别为：'小梯度优势','大梯度优势','灰度分布不均匀性','梯度分布不均匀性', % '能量','灰度平均','梯度平均','灰度方差','梯度方差','相关','灰度熵','梯度熵','混合熵','惯性','逆差距' % 灰度差分统计特征值： 对比度，熵，平均值,能量 % 特征数量 64个=16组位置算子,每组位置算子提取4个特征 % 灰度行程长度统计方法 % 提取特征：'强调短行程的逆差','强调长行程的逆差','灰度不均匀性','行程长度非均匀性','以行程表示的图像分数' % 三个方向，共获得15个特征 % 不变矩特征，三个颜色平面，每个颜色平面提取7个不变矩，共21个特征 % % 调用格式 [featureName]=featureNameMap(featureNO) % % 输入 featureNameMap(featureNO),featureNO为特征的编号 % 比如输入： featureNameMap(145),145即为当前的特征编号 % 输出： 'GrayCoOccurrenctMatrix-1X1-AngularSecondMoment' % '灰度共生矩阵-1X1-角二阶矩' % 表示灰度共生矩阵在方向(1,1)提取的'角二阶矩'这一特征 % % 2007.03.25 提取特征数量：208 % 2007-10-26 提取特征数量：391 % 增加特征种类： YIQ颜色特征、灰度-梯度共生矩阵纹理特征、灰度行程长度统计方法、不变矩特征 % 删除特征种类： Harr-like特征、傅立叶变换特征、以及边缘检测特征 % % 2007-10-26 柳锋 % function [featureName]=featureNameMap(featureNO) isRGB=1; isHSI=1; isGabor=1; isCoOcuMat=1; isGrayDiff=1; containFeature=[isRGB isHSI isGabor isCoOcuMat isGrayDiff]; rgbFeaNameEN={'RGB-R-Mean','RGB-G-Mean','RGB-B-Mean','RGB-R-Std','RGB-G-Std','RGB-B-Std',... 'RGB-R-Variance','RGB-G-Variance','RGB-B-Variance','RGB-RG-Correlation','RGB-GB-Correlation','RGB-RB-Correlation'}; hsiFeaNameEN={'HSI-H-Mean','HSI-S-Mean','HSI-I-Mean','HSI-H-Std','HSI-S-Std','HSI-I-Std'... 'HSI-H-Variance','HSI-S-Variance','HSI-I-Variance','HSI-HS-Correlation','HSI-SI-Correlation','HSI-HI-Correlation'}; ntscFeaNameEN={'YIQ-Y-Mean','YIQ-I-Mean','YIQ-Q-Mean','YIQ-Y-Std','YIQ-I-Std','YIQ-Q-Std'... 'YIQ-Y-Variance','YIQ-I-Variance','YIQ-Q-Variance','YIQ-YI-Correlation','YIQ-IQ-Correlation','YIQ-YQ-Correlation'}; rgbFeaNameCN={'RGB-R-均值','RGB-G-均值','RGB-B-均值','RGB-R-标准差','RGB-G-标准差','RGB-B-标准差'... 'RGB-R-方差','RGB-G-方差','RGB-B-方差','RGB-RG-相关性','RGB-GB-相关性','RGB-RB-相关性'}; hsiFeaNameCN={'HSI-H-均值','HSI-S-均值','HSI-I-均值','HSI-H-标准差','HSI-S-标准差','HSI-I-标准差'... 'HSI-H-方差','HSI-S-方差','HSI-I-方差','HSI-HS-相关性','HSI-SI-相关性','HSI-HI-相关性'}; ntscFeaNameCN={'YIQ-Y-均值','YIQ-I-均值','YIQ-Q-均值','YIQ-Y-标准差','YIQ-I-标准差','YIQ-Q-标准差'... 'YIQ-Y-方差','YIQ-I-方差','YIQ-Q-方差','YIQ-YI-相关性','YIQ-IQ-相关性','YIQ-YQ-相关性'}; % gabor 特征名称 fList=[0.5 0.25 0.125 0.1]; thetaList=((0:7)/8).*pi; gaborOrientation=(thetaList.*180)/pi; gaborFrequency=fList; gaborScale=1./(2.*gaborFrequency); counter=1; gaborFeaNameLen=2; gaborFeaNameEN=cell(1,length(gaborScale)*length(gaborOrientation)); gaborFeaNameCN=cell(1,length(gaborScale)*length(gaborOrientation)); gaborFeaNameBaseEN={'Mean','Std'}; gaborFeaNameBaseCN={'均值','标准差'}; for i=1:length(gaborScale) for j=1:length(gaborOrientation) for k=1:gaborFeaNameLen gaborFeaNameEN{counter}=strcat('GaborTransform-',num2str(gaborScale(i)),'X',num2str(gaborOrientation(j)),'-',gaborFeaNameBaseEN{k}); gaborFeaNameCN{counter}=strcat('Gabor变换-',num2str(gaborScale(i)),'X',num2str(gaborOrientation(j)),'-',gaborFeaNameBaseCN{k}); counter=counter+1; end end end %灰度共生矩阵特征名称 CoOcuMatFeaNameLen=11; CoOcuMatOffset=[0 1;0 5;1 0;5 0;1 1;1 -1;5 5;5 -5;2 5;2 -5;4 5;4 -5;5 2;5 -2;5 4;5 -4]; CoOcuMatFeaNameEN=cell(1,length(CoOcuMatOffset)*CoOcuMatFeaNameLen); CoOcuMatFeaNameCN=cell(1,length(CoOcuMatOffset)*CoOcuMatFeaNameLen); CoOCuMatFeaNameBaseEN={'AngularSecondMoment','Contrast','Correlation','DifferenceMoment','Homogeneity','SumMean',... 'SumVariance','SumEntropy','Entropy','DifferenceVariance','DifferenceEntropy'}; CoOCuMatFeaNameBaseCN={'角二阶矩','对比度','相关性','差分矩','逆差分矩（均匀性）','和平均值',... '和方差','和熵','熵','差方差','差熵'}; for i=1:length(CoOcuMatOffset) for j=1:CoOcuMatFeaNameLen CoOcuMatFeaNameEN{(i-1)*CoOcuMatFeaNameLen+j}=strcat('GrayCoOccurrenctMatrix-',num2str(CoOcuMatOffset(i,1)),'X',num2str(CoOcuMatOffset(i,2)),'-',CoOCuMatFeaNameBaseEN{j}); CoOcuMatFeaNameCN{(i-1)*CoOcuMatFeaNameLen+j}=strcat('灰度共生矩阵-',num2str(CoOcuMatOffset(i,1)),'X',num2str(CoOcuMatOffset(i,2)),'-',CoOCuMatFeaNameBaseCN{j}); end end %灰度-梯度 共生矩阵特征名称 GGLCMFeaNameLen=15; GGLCMFeaNameEN=cell(1,GGLCMFeaNameLen); GGLCMFeaNameCN=cell(1,GGLCMFeaNameLen); GGLCMFeaNameBaseEN={'SmallGradsDominance','BigGradsDominance','GrayAsymmetry','GradsAsymmetry','Energy','GrayMean','GradsMean',... 'GrayVariance','GradsVariance','Correlation','GrayEntropy','GradsEntropy','Entropy','Inertia','DifferMoment'}; GGLCMFeaNameBaseCN={'小梯度优势','大梯度优势','灰度分布不均匀性','梯度分布不均匀性','能量','灰度平均','梯度平均',... '灰度方差','梯度方差','相关','灰度熵','梯度熵','混合熵','惯性','逆差距'}; for i=1:GGLCMFeaNameLen GGLCMFeaNameEN{i}=strcat('GrayGradsCoOccurrenctMatrix-',GGLCMFeaNameBaseEN{i}); GGLCMFeaNameCN{i}=strcat('灰度-梯度共生矩阵-',GGLCMFeaNameBaseCN{i}); end %灰度差分统计 GrayDiffFeaNameLen=4; GrayDiffOffset=[0 1;0 3;0 5;1 0;3 0;5 0;1 1;2 2;4 4;5 5;1 3;3 1;2 4;4 2;3 5;5 3]; GrayDiffFeaNameEN=cell(1,length(GrayDiffOffset)*GrayDiffFeaNameLen); GrayDiffFeaNameCN=cell(1,length(GrayDiffOffset)*GrayDiffFeaNameLen); GrayDiffFeaNameBaseEN={'Contrast','Entropy','Mean','Energy'}; GrayDiffFeaNameBaseCN={'对比度','熵','均值','能量'}; for i=1:length(GrayDiffOffset) for j=1:GrayDiffFeaNameLen GrayDiffFeaNameEN{(i-1)*GrayDiffFeaNameLen+j}=strcat('GrayDifferenceStatistic-',num2str(GrayDiffOffset(i,1)),'X',num2str(GrayDiffOffset(i,2)),'-',GrayDiffFeaNameBaseEN{j}); GrayDiffFeaNameCN{(i-1)*GrayDiffFeaNameLen+j}=strcat('灰度差分统计-',num2str(GrayDiffOffset(i,1)),'X',num2str(GrayDiffOffset(i,2)),'-',GrayDiffFeaNameBaseCN{j}); end end % 灰度行程长度统计方法 GrayRunLengthFeaNameLen=5; thetaList=[0 pi/4 pi/2]; % 行程方向 grlmOrientation=(thetaList.*180)/pi; grlmFeaNameEN=cell(1,length(grlmOrientation)*GrayRunLengthFeaNameLen); grlmFeaNameCN=cell(1,length(grlmOrientation)*GrayRunLengthFeaNameLen); grlmFeaNameBaseEN={'ShortDifferMoment','LongDifferMoment','GrayAsymmetry','RunLengthAsymmetry','ImageFraction'}; grlmFeaNameBaseCN={'强调短行程的逆差','强调长行程的逆差','灰度不均匀性','行程长度非均匀性','以行程表示的图像分数'}; for i=1:length(grlmOrientation) for j=1:GrayRunLengthFeaNameLen grlmFeaNameEN{(i-1)*GrayRunLengthFeaNameLen+j}=strcat('GrayRunLengthStatistic-','Orientation:',num2str(grlmOrientation(i)),'-',grlmFeaNameBaseEN{j}); grlmFeaNameCN{