```
function output = getBiconcept( image )
%GETBICONCEPT Summary of this function goes here
% Detailed explanation goes here
% cd vlfeat-0.9.16\toolbox;
% run vl_setup;
% cd ..;
% cd ..;
if ~isdeployed
addpath 'vlfeat-0.9.16\toolbox';
run vl_setup;
end
load('model.mat');
load('model_2000.mat');
try
im = imread(image);
catch
disp('Failed to read image!');
output = -1;
return;
end
[w,h,c] = size(im);
if (w<10||h<h)
disp('Image too small!');
output = -1;
return;
end
if (c==1)
im=gray2rgb(im);
end
k1=getImageDescriptor(model, im);
k3=cat(2,k1);
gBOW= vl_homkermap(k3, 1, 'kchi2', 'gamma', .5) ;
%save([fullfile(outpath,outname) '-BOW.mat'],'gBOW');
%comingFile=image;
%extract gist feature
img = im;
img = rgb2gray(img);
imgsize=size(img);
lengthTemp=imgsize(1);
widthTemp=imgsize(2);
if lengthTemp>widthTemp
img=img(int32((lengthTemp-widthTemp)/2+1):int32((lengthTemp+widthTemp)/2),:);
else
img=img(:,int32((widthTemp-lengthTemp)/2+1):int32((lengthTemp+widthTemp)/2));
end
imgsize=size(img);
lengthTemp=imgsize(1);
widthTemp=imgsize(2);
img=img(1:min(lengthTemp,widthTemp),1:min(lengthTemp,widthTemp));
imgsize=size(img);
lengthTemp=imgsize(1);
widthTemp=imgsize(2);
% Parameters:
Nblocks = 4;
imageSize = min(lengthTemp,widthTemp);
orientationsPerScale = [8 8 4];
numberBlocks = 4;
% Precompute filter transfert functions (only need to do this one, unless image size is changes):
createGabor(orientationsPerScale, imageSize); % this shows the filters
G = createGabor(orientationsPerScale, imageSize);
% Computing gist requires 1) prefilter image, 2) filter image and collect
% output energies
output = prefilt(double(img), 4);
gGist = gistGabor(output, numberBlocks, G);
%save([fullfile(outpath,outname) '-Gist.mat'],'gGist');
% extract lbp features
img = im;
mapping=getmapping(8,'u2');
g=lbp(img,1,8,mapping,'h');
gLBP=g';
%save([fullfile(outpath,outname) '-LBP.mat'],'gLBP');
%%extract color hist feature
I = im;
nBins = 256;
rHist = imhist(I(:,:,1), nBins);
gHist = imhist(I(:,:,2), nBins);
bHist = imhist(I(:,:,3), nBins);
gColor=[rHist;gHist;bHist];
%save([fullfile(outpath,outname) '-Color.mat'],'gColor');
gBOW=normc(gBOW);
gGist=normc(gGist);
gLBP=normc(gLBP);
gColor=normc(gColor);
att = att_ext(gBOW', gLBP', gGist', gColor', model_all);
gAtt = att';
%save([fullfile(outpath,outname) '-Att.mat'],'gAtt');
gAtt=normc(gAtt);
gEarly=[gLBP' gBOW' gGist' gAtt' gColor'];
%save([fullfile(outpath,outname) '-Early.mat'],'gEarly');
output = 1;
load('classes.mat');
classnum = length(classes);
biconceptVector=zeros(classnum,1);
g=gEarly;
targetFile='testEarlyFusion';
label=1;
fileID = fopen(targetFile,'a+');
fprintf(fileID,'%d',label);
fprintf(fileID,' ');
[row,col]=size(g);
for j = 1:col
fprintf(fileID,'%d',j);
fprintf(fileID,':');
if(isnan(g(j)))
fprintf(fileID,'%f',0);
else
fprintf(fileID,'%f',g(j));
end
fprintf(fileID,' ');
end
fprintf(fileID,'\n');
fclose(fileID);
[label_descriptortest, inst_descriptortest] = libsvmread('testEarlyFusion');
for tempj=1:classnum
currentClass=classes{tempj};
try
load(['SVM\' currentClass '.mat']);
catch exception
end
[predict_label, accuracy, dec_values] = predict(label_descriptortest, inst_descriptortest, modelfusion,'-b 1');
biconceptVector(tempj,1)=dec_values(1);
end
%biconceptVector = normc(biconceptVector);
save(['result\' image(1:end-4) '-biconcept.mat'],'biconceptVector');
dlmwrite(['result\' image(1:end-4) '-biconcept.txt'],biconceptVector, ' ');
delete(targetFile)
end```这段代码为什么有标红的地方，导致运行错误“出错 getBiconcept (line 146)
[predict_label, accuracy, dec_values] = predict(label_descriptortest, inst_descriptortest, modelfusion,'-b 1');”，为什么今天上午还能正常运行，但是现在不可以了
本文介绍如何在MATLAB中配置并使用LIBSVM工具箱。首先确保已安装MATLAB的C++编译器，然后下载LIBSVM并解压缩至所需位置，通过设置路径并运行make命令来完成配置。
扫一扫
898
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
