Gabor变换的opencv实现

2024-02-10 20:18
文章标签 实现 opencv 变换 gabor

本文主要是介绍Gabor变换的opencv实现,希望对大家解决编程问题提供一定的参考价值,需要的开发者们随着小编来一起学习吧!

一直在做人脸识别方向,想用一下Gabor滤波器做一下Gabor人脸。从网上看到 Mian Zhou. Thesis. Gabor-Boosting Face Recognition的Gabor代码,也是基于Opencv的但是,代码是基于C结构的,用起来感觉不方便,就花一天的时间,修改成了C++结构的。
/*
Copyright (C) 2014 by Fangqi Su
Reference: Mian Zhou. Thesis. Gabor-Boosting Face Recognition
*/
#ifndef GABORFLITER_H
#define GABORFLITER_H
#include<iostream>
#include<opencv2\opencv.hpp>
#include<highgui.h>
#define CV_GABOR_REAL 1
#define CV_GABOR_IMAG 2
#define CV_GABOR_MAG 3
#define CV_GABOR_PHASE 4
using namespace cv;
class Gabor
{public:Gabor();Gabor(int iMu,int iNu,double dSigma);Gabor(int iMu,int iNu,double dSigma,double dF);Gabor(double dPhi,int iNu);Gabor(double dPhi,int iNu,double dSigma);Gabor(double dPhi,int iNu,double dSigma,double dF);Gabor(int iMu,int iNu);bool IsInit();bool IsKernelCreate();long mask_width();void Init(int iMu,int iNu,double dSigma,double dF);void Init(double dPhi,int iNu,double dSigma,double dF);void output_file(const char *filename,int Type);Mat get_matrix(int Type);Mat get_image(int Type);void show(int Type);void conv_img(Mat &src,Mat&dst,int Type);long Gabor::get_mask_width();protected:int KernelSize;double Sigma;double F;double Kmax;double K;double Phi;bool bInitialised;bool bKernel;long Width;Mat Imag;Mat Real;private:void creat_kernel();};#endif
/*
Copyright (C) 2014 by Fangqi Su
Reference: Mian Zhou. Thesis. Gabor-Boosting Face Recognition
*/
#include"GaborFliter.h"Gabor::Gabor()
{}
/*
fn Gabor::Gabor(int iMu,int iNu,double dSigma);
Constuct a gabor
Parameters:
iMu		The orientation iMu*PI/8,
iNu 		The scale,
dSigma 		The sigma value of Gabor,Create a gabor with a orientation iMu*PI/8, a scale iNu, and a sigma value dSigma. The spatial frequence (F) is set to sqrt(2) defaultly. It calls Init() to generate parameters and kernels.
*/
Gabor::Gabor(int iMu,int iNu,double dSigma)
{F=sqrt(2.0);Init(iMu,iNu,dSigma,F);
}
/*!
\fn Gabor::Gabor(int iMu, int iNu, double dSigma, double dF)
Construct a gaborParameters:
iMu		    The orientation iMu*PI/8
iNu 		The scale
dSigma 		The sigma value of Gabor
dF		    The spatial frequency Returns:
NoneCreate a gabor with a orientation iMu*PI/8, a scale iNu, a sigma value dSigma, and a spatial frequence dF. It calls Init() to generate parameters and kernels.
*/
Gabor::Gabor(int iMu,int iNu,double dSigma,double dF)
{Init(iMu,iNu,dSigma,dF);
}
/*!
\fn Gabor::Gabor(double dPhi, int iNu)
Construct a gaborParameters:
dPhi		The orientation in arc
iNu 		The scaleReturns:
NoneCreate a gabor with a orientation dPhi, and with a scale iNu. The sigma (Sigma) and the spatial frequence (F) are set to 2*PI and sqrt(2) defaultly. It calls Init() to generate parameters and kernels.
*/
Gabor::Gabor(double dPhi,int iNu)
{Sigma=2*CV_PI;F=sqrt(2.0);Init(dPhi,iNu,Sigma,F);
}
/*!
\fn Gabor::Gabor(double dPhi, int iNu, double dSigma)
Construct a gaborParameters:
dPhi		The orientation in arc
iNu 		The scale
dSigma		The sigma value of GaborReturns:
NoneCreate a gabor with a orientation dPhi, a scale iNu, and a sigma value dSigma. The spatial frequence (F) is set to sqrt(2) defaultly. It calls Init() to generate parameters and kernels.
*/
Gabor::Gabor(double dPhi,int iNu,double dSigma)
{F=sqrt(2.0);Init(dPhi,iNu,dSigma,F);
}
/*!
\fn Gabor::Gabor(double dPhi, int iNu, double dSigma, double dF)
Construct a gaborParameters:
dPhi		The orientation in arc
iNu 		The scale
dSigma 		The sigma value of Gabor
dF		The spatial frequency Returns:
NoneCreate a gabor with a orientation dPhi, a scale iNu, a sigma value dSigma, and a spatial frequence dF. It calls Init() to generate parameters and kernels.
*/
Gabor::Gabor(double dPhi,int iNu,double dSigma,double dF)
{Init(dPhi,iNu,dSigma,dF);
}
/*!
\fn Gabor::IsInit()
Determine the gabor is initilised or notParameters:
NoneReturns:
a boolean value, TRUE is initilised or FALSE is non-initilised.Determine whether the gabor has been initlized - variables F, K, Kmax, Phi, Sigma are filled.
*/
bool Gabor::IsInit()
{return bInitialised;
}
/*!
\fn Gabor::mask_width()
Give out the width of the maskParameters:
NoneReturns:
The long type show the width.Return the width of mask (should be NxN) by the value of Sigma and iNu.
*/
long Gabor::mask_width()
{long lWidth;if(IsInit()==false){perror("Error:Error: The Object has not been initilised in mask_width()!\n");return 0;}else {double dModSigma=Sigma/K;double dWidth=cvRound(dModSigma*6+1);if(fmod(dWidth,2.0)==0.0)dWidth++;lWidth=(long)dWidth;return lWidth;}
}
/*!
\fn Gabor::creat_kernel()
Create gabor kernelParameters:
NoneReturns:
NoneCreate 2 gabor kernels - REAL and IMAG, with an orientation and a scale 
*/
void Gabor::creat_kernel()
{if (IsInit() == false) {perror("Error: The Object has not been initilised in creat_kernel()!\n");}else {Mat mReal(Width,Width,CV_32F),mImag(Width,Width,CV_32F);/**************************** Gabor Function ****************************/ int x,y;double dReal;double dImag;double dTemp1,dTemp2,dTemp3;for(int i=0;i<Width;i++){for(int j=0;j<Width;j++){x=i-(Width-1)/2;y=j-(Width-1)/2;dTemp1 = (pow(K,2)/pow(Sigma,2))*exp(-(pow((double)x,2)+pow((double)y,2))*pow(K,2)/(2*pow(Sigma,2)));dTemp2 = cos(K*cos(Phi)*x + K*sin(Phi)*y) - exp(-(pow(Sigma,2)/2));dTemp3 = sin(K*cos(Phi)*x + K*sin(Phi)*y);dReal = dTemp1*dTemp2;dImag = dTemp1*dTemp3; mReal.at<float>(i,j)=dReal;mImag.at<float>(i,j)=dImag;}}/**************************** Gabor Function ****************************/bKernel=true;mReal.copyTo(Real);mImag.copyTo(Imag);}
}
/*!
\fn Gabor::get_image(int Type)
Get the speific type of image of GaborParameters:
Type		The Type of gabor kernel, e.g. REAL, IMAG, MAG, PHASE   Returns:
Pointer to image structure, or NULL on failure	Return an Image (gandalf image class) with a specific Type   "REAL"	"IMAG" "MAG" "PHASE"  
*/
Mat Gabor::get_image(int Type)
{if(IsKernelCreate() == false){ perror("Error: the Gabor kernel has not been created in get_image()!\n");return Mat();}Mat pImage(Width,Width,CV_32F,1);Mat newimage(Width,Width,CV_32FC1);Mat kernel(Width,Width,CV_32FC1);Mat re(Width,Width,CV_32FC1);Mat im(Width,Width,CV_32FC1);Mat ve;int rows=kernel.rows;int cols=kernel.cols;switch (Type){case 1://realReal.copyTo(kernel);kernel.copyTo(pImage);pImage=pImage.t();break;case 2://ImagImag.copyTo(kernel);kernel.copyTo(pImage);pImage=pImage.t();break;case 3://MagnitudeReal.copyTo(re);Imag.copyTo(im);pow(re,2,re);pow(im,2,im);pow(im+re,0.5,ve);ve.copyTo(pImage);pImage=pImage.t();break;default:break;}normalize(pImage,pImage,0,255,CV_MINMAX,NULL);pImage.convertTo(pImage,CV_8U);convertScaleAbs(pImage,newimage,1,0);newimage.convertTo(newimage,CV_8U);return newimage;
}
/*!
\fn Gabor::IsKernelCreate()
Determine the gabor kernel is created or notParameters:
NoneReturns:
a boolean value, TRUE is created or FALSE is non-created.Determine whether a gabor kernel is created.
*/
bool Gabor::IsKernelCreate()
{return bKernel;
}
/*!
\fn Gabor::get_mask_width()
Reads the width of MaskParameters:
NoneReturns:
Pointer to long type width of mask.
*/
long Gabor::get_mask_width()
{return Width;
}
/*!
\fn Gabor::Init(int iMu, int iNu, double dSigma, double dF)
Initilize the.gaborParameters:
iMu 	The orientations which is iMu*PI.8
iNu 	The scale can be from -5 to infinit
dSigma 	The Sigma value of gabor, Normally set to 2*PI
dF 	The spatial frequence , normally is sqrt(2)Returns:Initilize the.gabor with the orientation iMu, the scale iNu, the sigma dSigma, the frequency dF, it will call the function creat_kernel(); So a gabor is created.
*/
void Gabor::Init(int iMu,int iNu,double dSigma,double dF)
{bInitialised=false;bKernel=false;F=dF;Sigma=dSigma;Kmax=CV_PI/2;// Absolute value of KK = Kmax / pow(F, (double)iNu);Phi = CV_PI*iMu/8;bInitialised = true;Width = mask_width();Real = Mat( Width, Width, CV_32FC1);Imag = Mat( Width, Width, CV_32FC1);creat_kernel();  }
/*!
\fn CvGabor::CvGabor(int iMu, int iNu)
*/
Gabor::Gabor(int iMu, int iNu)
{double dSigma = 2*CV_PI; F = sqrt(2.0);Init(iMu, iNu, dSigma, F);
}
/*!
\fn Gabor::Init(double dPhi, int iNu, double dSigma, double dF)
Initilize the.gaborParameters:
dPhi 	The orientations 
iNu 	The scale can be from -5 to infinit
dSigma 	The Sigma value of gabor, Normally set to 2*PI
dF 	The spatial frequence , normally is sqrt(2)Returns:
NoneInitilize the.gabor with the orientation dPhi, the scale iNu, the sigma dSigma, the frequency dF, it will call the function creat_kernel(); So a gabor is created.filename 	The name of the image file
file_format 	The format of the file, e.g. GAN_PNG_FORMAT
image 	The image structure to be written to the file
octrlstr 	Format-dependent control structure*/
void Gabor::Init(double dPhi, int iNu, double dSigma, double dF)
{bInitialised = false;bKernel = false;Sigma = dSigma;F = dF;Kmax =CV_PI/2;// Absolute value of KK = Kmax / pow(F, (double)iNu);Phi = dPhi;bInitialised = true;Width = mask_width();Real = Mat( Width, Width, CV_32FC1);Imag = Mat( Width, Width, CV_32FC1);creat_kernel();  
}
/*!
\fn Gabor::get_matrix(int Type)
Get a matrix by the type of kernelParameters:
Type		The type of kernel, e.g. REAL, IMAG, MAG, PHASEReturns:
Pointer to matrix structure, or NULL on failure.Return the gabor kernel.
*/
Mat Gabor::get_matrix(int Type)
{if (!IsKernelCreate()) {perror("Error: the gabor kernel has not been created!\n"); return Mat();}switch (Type){case CV_GABOR_REAL:return Real;break;case CV_GABOR_IMAG:return Imag;break;case CV_GABOR_MAG:return Mat();break;case CV_GABOR_PHASE:return Mat();break;}
}
/*!
\fn Gabor::output_file(const char *filename, Gan_ImageFileFormat file_format, int Type)
Writes a gabor kernel as an image file.Parameters:
filename 	The name of the image file
file_format 	The format of the file, e.g. GAN_PNG_FORMAT
Type		The Type of gabor kernel, e.g. REAL, IMAG, MAG, PHASE   
Returns:
NoneWrites an image from the provided image structure into the given file and the type of gabor kernel.
*/
void Gabor::output_file(const char *filename, int Type)
{Mat pImage;pImage = get_image(Type);if(pImage .data!= NULL){if( imwrite(filename, pImage )) printf("%s has been written successfully!\n", filename);else printf("Error: writting %s has failed!\n", filename);}else perror("Error: the image is empty in output_file()!\n"); }
/*!
\fn CvGabor::conv_img_a(IplImage *src, IplImage *dst, int Type)
*/
void Gabor::conv_img(Mat &src, Mat &dst, int Type)
{Mat mat;src.copyTo(mat);mat.convertTo(mat,CV_32FC1);mat=mat.t();Mat temp;Mat rmat(src.rows,src.cols,CV_32FC1);Mat imat(src.rows,src.cols,CV_32FC1);Mat kernel(Width,Width,CV_32FC1);switch (Type){case CV_GABOR_REAL:Real.copyTo(kernel);filter2D( mat, mat,mat.depth(),kernel, Point( (Width-1)/2, (Width-1)/2));break;case CV_GABOR_IMAG:Imag.copyTo(kernel);filter2D( mat, mat,mat.depth(),kernel, Point( (Width-1)/2, (Width-1)/2));break;case CV_GABOR_MAG:/* Real Response */Real.copyTo(kernel);filter2D( mat, rmat,rmat.depth(),kernel, Point( (Width-1)/2, (Width-1)/2));/* Imag Response */Imag.copyTo(kernel);filter2D( mat, imat,imat.depth(),kernel, Point( (Width-1)/2, (Width-1)/2));/* Magnitude response is the square root of the sum of the square of real response and imaginary response */pow(rmat,2,rmat);pow(imat,2,imat);add(rmat,imat,temp);pow(temp,0.5,temp);temp.copyTo(mat);break;case CV_GABOR_PHASE:break;}normalize(mat,mat,0,255,CV_MINMAX,NULL);mat.convertTo(mat,CV_8U);mat=mat.t();mat.copyTo(dst);
}

这篇关于Gabor变换的opencv实现的文章就介绍到这儿,希望我们推荐的文章对编程师们有所帮助!



http://www.chinasem.cn/article/697975

相关文章

Python使用FFmpeg实现高效音频格式转换工具

《Python使用FFmpeg实现高效音频格式转换工具》在数字音频处理领域,音频格式转换是一项基础但至关重要的功能,本文主要为大家介绍了Python如何使用FFmpeg实现强大功能的图形化音频转换工具... 目录概述功能详解软件效果展示主界面布局转换过程截图完成提示开发步骤详解1. 环境准备2. 项目功能结

SpringBoot使用ffmpeg实现视频压缩

《SpringBoot使用ffmpeg实现视频压缩》FFmpeg是一个开源的跨平台多媒体处理工具集,用于录制,转换,编辑和流式传输音频和视频,本文将使用ffmpeg实现视频压缩功能,有需要的可以参考... 目录核心功能1.格式转换2.编解码3.音视频处理4.流媒体支持5.滤镜(Filter)安装配置linu

在Spring Boot中实现HTTPS加密通信及常见问题排查

《在SpringBoot中实现HTTPS加密通信及常见问题排查》HTTPS是HTTP的安全版本,通过SSL/TLS协议为通讯提供加密、身份验证和数据完整性保护,下面通过本文给大家介绍在SpringB... 目录一、HTTPS核心原理1.加密流程概述2.加密技术组合二、证书体系详解1、证书类型对比2. 证书获

Druid连接池实现自定义数据库密码加解密功能

《Druid连接池实现自定义数据库密码加解密功能》在现代应用开发中,数据安全是至关重要的,本文将介绍如何在​​Druid​​连接池中实现自定义的数据库密码加解密功能,有需要的小伙伴可以参考一下... 目录1. 环境准备2. 密码加密算法的选择3. 自定义 ​​DruidDataSource​​ 的密码解密3

使用Python实现Windows系统垃圾清理

《使用Python实现Windows系统垃圾清理》Windows自带的磁盘清理工具功能有限,无法深度清理各类垃圾文件,所以本文为大家介绍了如何使用Python+PyQt5开发一个Windows系统垃圾... 目录一、开发背景与工具概述1.1 为什么需要专业清理工具1.2 工具设计理念二、工具核心功能解析2.

Java实现本地缓存的常用方案介绍

《Java实现本地缓存的常用方案介绍》本地缓存的代表技术主要有HashMap,GuavaCache,Caffeine和Encahche,这篇文章主要来和大家聊聊java利用这些技术分别实现本地缓存的方... 目录本地缓存实现方式HashMapConcurrentHashMapGuava CacheCaffe

SpringBoot整合Sa-Token实现RBAC权限模型的过程解析

《SpringBoot整合Sa-Token实现RBAC权限模型的过程解析》:本文主要介绍SpringBoot整合Sa-Token实现RBAC权限模型的过程解析,本文给大家介绍的非常详细,对大家的学... 目录前言一、基础概念1.1 RBAC模型核心概念1.2 Sa-Token核心功能1.3 环境准备二、表结

Python实现一键PDF转Word(附完整代码及详细步骤)

《Python实现一键PDF转Word(附完整代码及详细步骤)》pdf2docx是一个基于Python的第三方库,专门用于将PDF文件转换为可编辑的Word文档,下面我们就来看看如何通过pdf2doc... 目录引言:为什么需要PDF转Word一、pdf2docx介绍1. pdf2docx 是什么2. by

使用Python实现网页表格转换为markdown

《使用Python实现网页表格转换为markdown》在日常工作中,我们经常需要从网页上复制表格数据,并将其转换成Markdown格式,本文将使用Python编写一个网页表格转Markdown工具,需... 在日常工作中,我们经常需要从网页上复制表格数据,并将其转换成Markdown格式,以便在文档、邮件或

Python使用pynput模拟实现键盘自动输入工具

《Python使用pynput模拟实现键盘自动输入工具》在日常办公和软件开发中,我们经常需要处理大量重复的文本输入工作,所以本文就来和大家介绍一款使用Python的PyQt5库结合pynput键盘控制... 目录概述:当自动化遇上可视化功能全景图核心功能矩阵技术栈深度效果展示使用教程四步操作指南核心代码解析