OpenCV数据读写操作
1. 写yml矩阵
- 代码: 全选
#include "cxcore.h"
int main( int argc, char** argv )
{
CvMat* mat = cvCreateMat( 3, 3, CV_32F );
CvFileStorage* fs = cvOpenFileStorage( "example.yml", 0, CV_STORAGE_WRITE );
cvSetIdentity( mat );
cvWrite( fs, "A", mat, cvAttrList(0,0) );
cvReleaseFileStorage( &fs );
cvReleaseMat( &mat );
return 0;
}
Re: opencv 数据读写操作
2. 读yml矩阵
- 代码: 全选
#include <stdio.h>
#include "cxcore.h"
#include "highgui.h"
void main( )
{
CvMat* A1 = (CvMat*)cvLoad( "example.yml" );
for(int i = 0; i <3; i++ )
{
printf( "\n");
for(int j = 0; j < 3; j++ )
printf( "%f ", (float) cvGetReal2D( A1, i, j ));
}
printf( "\n\n");
}
Re: opencv 数据读写操作
3. 写xml矩阵
- 代码: 全选
#include <stdio.h>
#include "cxcore.h"
#include "highgui.h"
void main( )
{
double a[] = { 1, 0, 0, 0, 0,
0, 1, 0, 0, 0,
0, 0, 1, 0, 0,
0, 0, 0, 1, 0,
0, 0, 0, 0, 1 };
CvMat A1;
cvInitMatHeader( &A1, 5, 5, CV_64FC1, a, CV_AUTOSTEP );
cvSave( "my_matrix_test.xml", &A1 );
}
Re: opencv 数据读写操作
4. 读xml矩阵
- 代码: 全选
#include <stdio.h>
#include "cxcore.h"
#include "highgui.h"
void main( )
{
CvMat* A1 = (CvMat*)cvLoad( "my_matrix.xml" );
for(int i = 0; i <5; i++ )
{
printf( "\n");
for(int j = 0; j < 5; j++ )
printf( "%f ", (float) cvGetReal2D( A1, i, j ));
}
printf( "\n\n");
}
Re: opencv 数据读写操作
5. 写点列
- 代码: 全选
#include <stdio.h>
#include "highgui.h"
#include "cxcore.h"
void main( )
{
CvMemStorage* memstorage = cvCreateMemStorage(0);
CvSeq *up_seq = cvCreateSeq( CV_SEQ_ELTYPE_POINT,
sizeof(CvSeq),
sizeof(CvPoint),
memstorage);
for(int i=1; i<=100; i++)
{
// 产生单个点
CvPoint pt1 = cvPoint( i, i );
cvSeqPush(up_seq,&pt1);
cvSave("upimg_matched_point.yml", up_seq, NULL, "matched_points_up", cvAttrList(NULL,NULL));
}
}
Re: opencv 数据读写操作
6. 读点列
- 代码: 全选
#include <stdio.h>
#include "cxcore.h"
#include <stdio.h>
#include "cxcore.h"
int main( int argc, char** argv )
{
CvMemStorage* storage = cvCreateMemStorage(0);
CvSeq* seq;
CvSeqWriter writer;
CvSeqReader reader;
CvFileNode *node;
CvFileStorage* fs = cvOpenFileStorage( "upimg_matched_point.yml", 0, CV_STORAGE_READ );
node = cvGetFileNodeByName (fs, NULL, "upimg_matched_point");
seq = (CvSeq *) cvRead (fs, node);
cvStartReadSeq( seq, &reader, 0 );
for(int i = 0; i < 100; i++ )
{
CvPoint pt1;
#if 1
CV_READ_SEQ_ELEM( pt1, reader );
printf("%d %d is read\n", pt1.x, pt1.y );
#else
printf("%d is read\n", *(int*)reader.ptr );
CV_NEXT_SEQ_ELEM( seq->elem_size, reader );
#endif
}
cvReleaseFileStorage( &fs );
return 0;
}
Re: opencv 数据读写操作
7. 写图像xml数据
- 代码: 全选
#include <stdio.h>
#include "cv.h"
#include "cxcore.h"
#include "highgui.h"
void main( )
{
IplImage *Image1,*Image2;
int height,width,step,channels;
int i, j, k, i1;
uchar *data;
CvScalar Scalar1;
Image1=cvLoadImage("lena.jpg", 1);
height = Image1->height;
width = Image1->width;
step = Image1->widthStep;
channels = Image1->nChannels;
data = (uchar *)Image1->imageData;
cvNamedWindow("rawimage",1);
cvShowImage("rawimage", Image1);
cvWaitKey(0);
CvMat *M=cvCreateMat(width, height, CV_32FC3);
for(i=0; i<width; i++)
for(j=0; j<height; j++)
{
Scalar1=cvGet2D(Image1, i, j);
cvSet2D(M, i, j, Scalar1);
}
CvFileStorage* fs = cvOpenFileStorage("lena.xml", 0, CV_STORAGE_WRITE);
cvWriteInt( fs, "frame_count", 1 );
cvStartWriteStruct( fs, "frame_size", CV_NODE_SEQ);
cvWriteInt( fs, 0, width );
cvWriteInt( fs, 0, height );
cvEndWriteStruct(fs);
cvWrite( fs, "lena", M );
cvReleaseFileStorage( &fs );
cvReleaseImage( &Image1 );
}
Re: opencv 数据读写操作
8. 读图像xml数据
- 代码: 全选
#include <stdio.h>
#include "cv.h"
#include "highgui.h"
void main( )
{
IplImage *Image1;
int i, j;
CvFileNode *node;
CvScalar Scalar1;
CvFileStorage* fs = cvOpenFileStorage("lena.xml", 0, CV_STORAGE_READ);
int frame_count = cvReadIntByName( fs, 0, "frame_count", 10 /* default value */ );
CvSeq* s = cvGetFileNodeByName(fs, 0, "frame_size")->data.seq;
int width = cvReadInt( (CvFileNode*)cvGetSeqElem(s, 0) );
int height = cvReadInt( (CvFileNode*)cvGetSeqElem(s, 1) );
node = cvGetFileNodeByName(fs, NULL, "lena");
CvMat *M1=cvCreateMatHeader(width, height, CV_32FC3);
M1 = (CvMat *)cvRead( fs, cvGetFileNodeByName(fs, NULL, "lena"));
cvReleaseFileStorage( &fs );
Image1 = cvCreateImage( cvSize(width, height), 8, 3 );
for(i=0; i<width; i++)
for(j=0; j<height; j++)
{
Scalar1=cvGet2D(M1, i, j);
cvSet2D(Image1, i, j, Scalar1);
}
cvNamedWindow("rawimage",1);
cvShowImage("rawimage", Image1);
cvWaitKey(0);
cvReleaseFileStorage( &fs );
cvReleaseImage( &Image1 );
}
Re: opencv 数据读写操作
9. 由数据产生图像
- 代码: 全选
#include "cv.h"
#include "highgui.h"
#include <math.h>
#include <stdio.h>
#include <stdlib.h>
void main()
{
int step,channels;
uchar *data;
// 创建空白图板
int width=301, height=301;
IplImage* image = cvCreateImage( cvSize(width,height), 8, 1 );
// 获取图像格式
step = image->widthStep;
channels = image->nChannels;
data = (uchar *)image->imageData;
// 产生图像数据
for(int i=0; i<height; i++)
for(int j=0; j<width; j++)
{
// for color image channels changes from 0 to 2
img_dat= function[(i-centre_i, j-centre_j)];
// your subroutine here.
data[i*step+j*channels]=max(0, min(255, (int)img_dat));
}
cvNamedWindow( "My_img", 1 );
cvShowImage( "My_img", image );
cvSaveImage( "My_img.bmp", image );
cvReleaseImage(&image );
cvWaitKey(0);
}
点操作是图像处理的底层算法，以图像加噪为例
1. 高斯噪声和均匀分布噪声
- 代码: 全选
// 初始化随机数发生器
CvRNG rng = cvRNG(-1);
// 产生高斯噪声数据
cvRandArr( &rng, noise, CV_RAND_NORMAL, cvScalarAll(0), cvScalarAll(sigma) );
// 产生均匀分布噪声数据
cvRandArr( &rng, noise, CV_RAND_UNI, cvScalarAll(-sigma), cvScalarAll(sigma) );
Re: opencv 数据读写操作
2. 添加泊松噪声
用CV_IMAGE_ELEM的速度是cvGet2D() + cvSet2D() 的10倍。
- 代码: 全选
// 设定噪声图像模板
noise=cvCreateImage( size, IPL_DEPTH_8U, channels );
srand( time(NULL) ); // 设置随机数种子的
for(i=0; i<height; i++)
for(j=0; j<width; j++)
{
for(k=0; k<channels; k++)
{
i1=rand();
x1=(float)i1/32767.0;
x2= numda*sqrt(-2* log(1-x1));
CV_IMAGE_ELEM(noise, unsigned char, i, j*3+k) = x2;
}
}
// 图像加噪操作
for(i=0; i<height; i++)
for(j=0; j<width; j++)
{
for( k=0; k<channels; k++)
{
CV_IMAGE_ELEM(image, unsigned char, i, j*3+k) = max(0, min(255, CV_IMAGE_ELEM(image, unsigned char, i, j*3+k)
+ CV_IMAGE_ELEM(noise, unsigned char, i, j*3+k)));
}
}
用CV_IMAGE_ELEM的速度是cvGet2D() + cvSet2D() 的10倍。
Re: opencv 数据读写操作
3. 添加椒盐噪声
- 代码: 全选
float psalt=0.1; //probability salt
float ppepper=0.2; //probability pepper
for(i=0; i<height; i++)
for(j=0; j<width; j++)
{
float valsalt=(float)rand()/65535;
float valpepper=(float)rand()/65535;
s=cvGet2D(image1, i, j);
for(k=0; k<channels; k++)
{
if (valsalt<=psalt && valsalt>valpepper)
s.val[k]=255;
else if(valpepper<=ppepper && valpepper>=valsalt)
s.val[k]=0;
}
cvSet2D(image1, i, j, s);
}
cvAdd(image1, noise, image2);
Re: opencv 数据读写操作
4. 添加柏林噪声
perlin.h
产生噪声数据
perlin.h
- 代码: 全选
// Perlin.h
#include <stdlib.h>
#include <stdio.h>
#include <math.h>
float persistence = 0.55f; // 可调参数1，可由editbox输入
int Number_Of_Octaves = 3; // 可调参数2，可由editbox输入
/* 一个噪声发生器 */
float Noise1(int x, int y)
{
int n = x + y * 57;
n = (n<<13) ^ n;
return ( 1.0f - ( (n * (n * n * 15731 + 789221) + 1376312589) & 0x7fffffff) / 1073741824.0f);
}
/* 一个光滑噪声发生器 */
float SmoothNoise_1(int x, int y)
{
float corners = ( Noise1(x-1, y-1)+Noise1(x+1, y-1)+Noise1(x-1, y+1)+Noise1(x+1, y+1) ) / 16.0f;
float sides = ( Noise1(x-1, y) +Noise1(x+1, y) +Noise1(x, y-1) +Noise1(x, y+1) ) / 8.0f;
float center = Noise1(x, y) / 4.0f;
return corners + sides + center;
}
/* 插值函数 */
float Cosine_Interpolate(float a, float b, float x)
{
double ft = x * 3.1415927;
double f = (1 - cos(ft)) * 0.5f;
return (float)(a*(1-f) + b*f);
}
/* 插值噪声发生器 */
float InterpolatedNoise_1(float x, float y)
{
int integer_X = (int)floor(x+0.5);
float fractional_X = x - integer_X;
int integer_Y = (int)floor(y+0.5);
float fractional_Y = y - integer_Y;
float v1 = SmoothNoise_1(integer_X, integer_Y);
float v2 = SmoothNoise_1(integer_X + 1, integer_Y);
float v3 = SmoothNoise_1(integer_X, integer_Y + 1);
float v4 = SmoothNoise_1(integer_X + 1, integer_Y + 1);
float i1 = Cosine_Interpolate(v1 , v2 , fractional_X);
float i2 = Cosine_Interpolate(v3 , v4 , fractional_X);
return Cosine_Interpolate(i1 , i2 , fractional_Y);
}
/* 最终的PERLIN NOISE */
float PerlinNoise_2D(float x, float y)
{
float total = 0.0f;
float p = persistence;
int n = Number_Of_Octaves - 1;
int ii;
for(ii=0;ii<=n;ii++)
{
float frequency = (float) pow((float)2,ii);
float amplitude = (float) pow(p,ii);
total = total + InterpolatedNoise_1(x * frequency, y * frequency) * amplitude;
}
return total;
}
float myperlin( int x, int y )
{
int ii, jj;
float ynoise=0;
for(ii=0; ii< x; ii++)
for(jj=0; jj< y; jj++)
{
ynoise=PerlinNoise_2D((float)ii, (float)jj);
ynoise= (float)(1+ynoise)*127.0;
}
return ynoise;
}
产生噪声数据
- 代码: 全选
// 产生噪声
CvMat *M=cvCreateMat(width, height, CV_32FC1); // 不可用CV_8C1
IplImage* dst=cvCreateImage(size, img->depth, img->nChannels);
for(i=0; i<height; i++)
for(j=0; j<width; j++)
{
x1= myperlin(i, j);
cvmSet(M, j, i, x1 );
}
我在vc下可以做得很快，但在opencv下只能处理64X64
5. 1/f噪声
下面给出一维1/f噪声数组产生程序（1/f噪声的谱方法）
下面给出一维1/f噪声数组产生程序（1/f噪声的谱方法）
- 代码:
function sig=FractRnd(beta, n, HuaTu)
%sig=FractRnd(beta, n)
%To generate 1/f^betta noise
%INPUT
% beta is the exponent, a number
% n is the number of synthetic data points
%OUTPUT
% sig is an n-points array
%
if nargin==2
HuaTu=0;
end
len_beta=length(beta);
phi_n=2*pi*rand(1,n);
f=[1:n].^(-beta/2).*exp(i*phi_n(1:n));
sig=real(ifft(f));
sig=(sig-mean(sig))./std(sig);
if HuaTu
%To verify the scaling behavior of power spectral density:
% P(f) ~ 1/f^beta
figure(HuaTu)
F=fft(sig);
P=(abs(F)).^2;
P([1,round(end/2):end])=[];
% plot(log(1:length(P)),log(P))
loglog(1:length(P),P,'k')
xlabel('Frequency');
ylabel('Power spectrum')
% pause
end
在程序中可以通过vc调用matlab实现
