GitHub - zzsi/EMScanSupression: EM clustering under sparse coding principle

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 15 Commits
positiveImage		positiveImage
ApplyFilterfftSame.m		ApplyFilterfftSame.m
CorrFilter.m		CorrFilter.m
ExponentialModel.m		ExponentialModel.m
GaborFilter.m		GaborFilter.m
GenerateHtml.m		GenerateHtml.m
MakeFilter.m		MakeFilter.m
ParameterCodeImage.m		ParameterCodeImage.m
README		README
SUM1MAX1.m		SUM1MAX1.m
StartFromHere.m		StartFromHere.m
ZIPproject.m		ZIPproject.m
displayActivations.m		displayActivations.m
displayImages.m		displayImages.m
displayMatchedTemplate.m		displayMatchedTemplate.m
displayTemplate.m		displayTemplate.m
drawGaborSymbol.m		drawGaborSymbol.m
encode.m		encode.m
inscribe.m		inscribe.m
mexc_ComputeMAX1.cpp		mexc_ComputeMAX1.cpp
mexc_ComputeMAX2.cpp		mexc_ComputeMAX2.cpp
mexc_ComputeMAX2MP.cpp		mexc_ComputeMAX2MP.cpp
mexc_ComputeSUM2.cpp		mexc_ComputeSUM2.cpp
mexc_CropInstance.cpp		mexc_CropInstance.cpp
mexc_Histogram.cpp		mexc_Histogram.cpp
mexc_LocalNormalize.cpp		mexc_LocalNormalize.cpp
mexc_LocalNormalizeDouble.cpp		mexc_LocalNormalizeDouble.cpp
mexc_SharedSketch.cpp		mexc_SharedSketch.cpp
mexc_Sigmoid.cpp		mexc_Sigmoid.cpp
mexc_TemplateAffineTransform.cpp		mexc_TemplateAffineTransform.cpp
storedExponentialModel1.mat		storedExponentialModel1.mat
test.m		test.m

Repository files navigation

This project is about learning visual words as deformable templates, using sparse coding principle.
It is achieved by sparse EM clustering on a large number of image patches collected from large images.
Sparse coding is used in the E step.

Run StartFromHere.m in Matlab, and monitor intermediate results in the folder ./output/ .

Then run GenerateHtml.m. This will generate a HTML page to document the experiment results. Please look
into the document/ folder.