VIBE算法源码.docx

VIBE算法源码.docx

《VIBE算法源码.docx》由会员分享，可在线阅读，更多相关《VIBE算法源码.docx（49页珍藏版）》请在冰点文库上搜索。

VIBE算法源码

VIBE算法源码

这是源代码，经过稍微的修改，可完美运行。

测试环境为vs2015+opencv3.4.6

无任何预处理，直接运行效果图：

vibe-background-sequential.h

/**

@filevibe-background-sequential.h

@briefInterfacefortheViBelibrary

@authorMarcVanDroogenbroeck

@dateJuly2014

@details

Fulldocumentationisavailableonlineat:

http:

//www.ulg.ac.be/telecom/research/vibe/doc

Alltechnicaldetailsareavailableinthefollowingpaper:

O.BarnichandM.VanDroogenbroeck.ViBe:

Auniversalbackgroundsubtractionalgorithmforvideosequences.IEEETransactionsonImageProcessing,20（6）:

1709-1724,June2011.

\verbatim

BiBTeXinformation

@article{Barnich2011ViBe,

title={{ViBe}:

Auniversalbackgroundsubtractionalgorithmforvideosequences},

author={O.BarnichandM.{VanDroogenbroeck}},

journal={IEEETransactionsonImageProcessing},

volume={20},

number={6},

pages={1709-1724},

month={June},

year={2011},

keywords={ViBe,Background,Backgroundsubtraction,Segmentation,Motion,Motiondetection},

pdf={http:

//orbi.ulg.ac.be/bitstream/2268/145853/1/Barnich2011ViBe.pdf},

doi={10.1109/TIP.2010.2101613},

url={

}

\endverbatim

See

\citeBarnich2011ViBe

*/

#ifndef_VIBE_SEQUENTIAL_H_

#define_VIBE_SEQUENTIAL_H_

#ifdef__cplusplus

extern"C"

{

#endif

#include

#include

#include

#include

#defineCOLOR_BACKGROUND0/*!

#defineCOLOR_FOREGROUND255/*!

/**

*\typedefstructvibeModel_Sequential_t

*\briefDatastructureforthebackgroundsubtractionmodel.

*

*Thisdatastructurecontainsthebackgroundmodelaswellassomeparamatersvalue.

*Thecodeisdesignedtohidealltheimplementationdetailstotheusertoeaseitsuse.

*/

typedefstructvibeModel_SequentialvibeModel_Sequential_t;

/**

*Allocationofanewdatastructurewherethebackgroundmodelwillbestored.

*Pleasenotethatthisfunctiononlycreatesthestructuretohostthedata.

*Thisdatastructureswillonlybefilledwithacallto\reflibvibeModel_Sequential_AllocInit_8u_C1R.

*

*\resultApointertoanewlyallocated\refvibeModel_Sequential_t

*structure,orNULLinthecaseofanerror.

*/

vibeModel_Sequential_t*libvibeModel_Sequential_New（）;

/**

*ViBeusesseveralparameters.

*Youcanprintandchangesomeofthemifyouwant.However,default

*valueshouldmeetyourneedsformostvideos.

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@return

*/

uint32_tlibvibeModel_Sequential_PrintParameters（constvibeModel_Sequential_t*model）;

/**

*Setter.

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@paramnumberOfSamples

*@return

*/

int32_tlibvibeModel_Sequential_SetNumberOfSamples（

vibeModel_Sequential_t*model,

constuint32_tnumberOfSamples

）;

/**

*Setter.

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@return

*/

uint32_tlibvibeModel_Sequential_GetNumberOfSamples（constvibeModel_Sequential_t*model）;

/**

*Setter.

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@parammatchingThreshold

*@return

*/

int32_tlibvibeModel_Sequential_SetMatchingThreshold（

vibeModel_Sequential_t*model,

constuint32_tmatchingThreshold

）;

/**

*Setter.

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@return

*/

uint32_tlibvibeModel_Sequential_GetMatchingThreshold（constvibeModel_Sequential_t*model）;

/**

*Setter.

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@parammatchingNumber

*@return

*/

int32_tlibvibeModel_Sequential_SetMatchingNumber（

vibeModel_Sequential_t*model,

constuint32_tmatchingNumber

）;

/**

*Setter.

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@paramupdateFactorNewvaluefortheupdatefactor.Pleasenotethattheupdatefactoristobeunderstoodasaprobabilityofupdating.Morespecifically,anupdatefactorof16meansthat1outofevery16backgroundpixelsisupdated.Likewise,anupdatefactorof1meansthateverybackgroundpixelisupdated.

*@return

*/

int32_tlibvibeModel_Sequential_SetUpdateFactor（

vibeModel_Sequential_t*model,

constuint32_tupdateFactor

）;

/**

*Getter.

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@return

*/

uint32_tlibvibeModel_Sequential_GetMatchingNumber（constvibeModel_Sequential_t*model）;

/**

*Getter.

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@return

*/

uint32_tlibvibeModel_Sequential_GetUpdateFactor（constvibeModel_Sequential_t*model）;

/**

*\briefFreesallthememoryusedbythemodelanddeallocatesthestructure.

*

*Thisfunctionfreesallthememoryallocatedby\reflibvibeModel_SequentialNewand

*\reflibvibeModel_Sequential_AllocInit_8u_C1Ror\reflibvibeModel_Sequential_AllocInit_8u_C3R.

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@return

*/

int32_tlibvibeModel_Sequential_Free（vibeModel_Sequential_t*model）;

/**

*Thetwofollowingfunctionsallocatetherequiredmemoryaccordingtothe

*modelparametersandthedimensionsoftheinputimages.

*Youmustusethe"C1R"functionforgrayscaleimagesandthe"C3R"forcolor

*images.

*These2functionsalsoinitializethebackgroundmodelusingthecontent

*of*image_datawhichisthepixelbufferofthefirstimageofyourstream.

*/

//-------------------------Singlechannelimages----------------------------

/**

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@paramimage_data

*@paramwidth

*@paramheight

*@return

*/

int32_tlibvibeModel_Sequential_AllocInit_8u_C1R（

vibeModel_Sequential_t*model,

constuint8_t*image_data,

constuint32_twidth,

constuint32_theight

）;

/*These2functionsperform2operations:

*-theyclassifythepixels*image_datausingtheprovidedmodelandstore

*theresultsin*segmentation_map.

*-theyupdate*modelaccordingtotheseresultsandthecontentof

**image_data.

*Youmustusethe"C1R"functionforgrayscaleimagesandthe"C3R"forcolor

*images.

*/

/**

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@paramimage_data

*@paramsegmentation_map

*@return

*/

int32_tlibvibeModel_Sequential_Segmentation_8u_C1R（

vibeModel_Sequential_t*model,

constuint8_t*image_data,

uint8_t*segmentation_map

）;

/**

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@paramimage_data

*@paramupdating_mask

*@return

*/

int32_tlibvibeModel_Sequential_Update_8u_C1R（

vibeModel_Sequential_t*model,

constuint8_t*image_data,

uint8_t*updating_mask

）;

//-------------------------Threechannelimages-----------------------------

/**

*Thepixelvaluesofcolorimagesarearrangedinthefollowingorder

*RGBRGBRGB...（orHSVHSVHSVHSVHSVHSV...）

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@paramimage_data

*@paramwidth

*@paramheight

*@return

*/

int32_tlibvibeModel_Sequential_AllocInit_8u_C3R（

vibeModel_Sequential_t*model,

constuint8_t*image_data,

constuint32_twidth,

constuint32_theight

）;

/*These2functionsperform2operations:

*-theyclassifythepixels*image_datausingtheprovidedmodelandstore

*theresultsin*segmentation_map.

*-theyupdate*modelaccordingtotheseresultsandthecontentof

**image_data.

*Youmustusethe"C1R"functionforgrayscaleimagesandthe"C3R"forcolor

*images.

*/

/**

*Thepixelvaluesofcolorimagesarearrangedinthefollowingorder

*RGBRGBRGB...（orHSVHSVHSVHSVHSVHSV...）

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@paramimage_data

*@paramsegmentation_map

*@return

*/

int32_tlibvibeModel_Sequential_Segmentation_8u_C3R（

vibeModel_Sequential_t*model,

constuint8_t*image_data,

uint8_t*segmentation_map

）;

/**

*Thepixelvaluesofcolorimagesarearrangedinthefollowingorder

*RGBRGBRGB...（orHSVHSVHSVHSVHSVHSV...）

*

*@parammodelThedatastructurewithViBe'sbackgroundsubtractionmodelandparameters.

*@paramimage_data

*@paramupdating_mask

*@return

*/

int32_tlibvibeModel_Sequential_Update_8u_C3R（

vibeModel_Sequential_t*model,

constuint8_t*image_data,

uint8_t*updating_mask

）;

#ifdef__cplusplus

}

#endif

#endif

vibe-background-sequential.cpp

/**

@filevibe-background-sequential.c

@briefImplementationofvibe-background-sequential.h

@authorMarcVanDroogenbroeck

@dateMay2014

*/

/*

Thereareafewimplementation"tricks"[byMarcVanDroogenbroeck].

PleasenotethatTHEYDON'TAFFECTTHEBEHAVIOROFViBebuthelpspeedinguptheimplementation;theyaredescribedhereafter.

1.Swappingvaluesinordertobringbestmatchingcandidatesinthefirstplacesofthememory

TheideaisthatViBeonlyneedstwomatchestoclassifyapixelintothebackground;tosomeextent,the18otherpixels（outofN=20pixels）arethenuseless.Inaddition,mostofthetime,therearenobigchangesbetweenpixelslocatedatthesameplacefromsuccessiveimages（90%ofthepixelsareinthebackgroundonaverage）.Ther