opencv/modules/objdetect/src/featurepyramid.cpp

578 lines
18 KiB
C++
Raw Normal View History

#include "precomp.hpp"
#include "_latentsvm.h"
#include "_lsvm_resizeimg.h"
#ifndef max
#define max(a,b) (((a) > (b)) ? (a) : (b))
#endif
#ifndef min
#define min(a,b) (((a) < (b)) ? (a) : (b))
#endif
2010-10-21 20:30:57 +02:00
static inline int sign(float r){
if(r > 0.0001f) return 1;
if(r < -0.0001f) return -1;
return 0;
}
/*
// Getting feature map for the selected subimage
//
// API
// int getFeatureMaps(const IplImage * image, const int k, featureMap **map);
// INPUT
// image - selected subimage
// k - size of cells
// OUTPUT
// map - feature map
// RESULT
// Error status
*/
int getFeatureMaps_dp(const IplImage * image,const int k, CvLSVMFeatureMap **map)
{
int sizeX, sizeY;
int p, px, strsz;
int height, width, channels;
int i, j, kk, c, ii, jj, d;
float * datadx, * datady;
float tmp, x, y, tx, ty;
IplImage * dx, * dy;
int *nearest_x, *nearest_y;
float *w, a_x, b_x;
float kernel[3] = {-1.f, 0.f, 1.f};
CvMat kernel_dx = cvMat(1, 3, CV_32F, kernel);
CvMat kernel_dy = cvMat(3, 1, CV_32F, kernel);
float * r;
int * alfa;
float boundary_x[CNTPARTION+1];
float boundary_y[CNTPARTION+1];
float max, tmp_scal;
int maxi;
height = image->height;
width = image->width ;
channels = image->nChannels;
dx = cvCreateImage(cvSize(image->width , image->height) , IPL_DEPTH_32F , 3);
dy = cvCreateImage(cvSize(image->width , image->height) , IPL_DEPTH_32F , 3);
sizeX = width / k;
sizeY = height / k;
px = CNTPARTION + 2 * CNTPARTION; // <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> <20> <20><> <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
p = px;
strsz = sizeX * p;
allocFeatureMapObject(map, sizeX, sizeY, p, px);
cvFilter2D(image, dx, &kernel_dx, cvPoint(-1, 0));
cvFilter2D(image, dy, &kernel_dy, cvPoint(0, -1));
for(i = 0; i <= CNTPARTION; i++)
{
boundary_x[i] = cosf((((float)i) * (((float)PI) / (float) (CNTPARTION))));
boundary_y[i] = sinf((((float)i) * (((float)PI) / (float) (CNTPARTION))));
}/*for(i = 0; i <= CNTPARTION; i++) */
r = (float *)malloc( sizeof(float) * (width * height));
alfa = (int *)malloc( sizeof(int ) * (width * height * 2));
for(j = 1; j < height-1; j++)
{
datadx = (float*)(dx->imageData + dx->widthStep *j);
datady = (float*)(dy->imageData + dy->widthStep *j);
for(i = 1; i < width-1; i++)
{
c = 0;
x = (datadx[i*channels+c]);
y = (datady[i*channels+c]);
r[j * width + i] =sqrtf(x*x + y*y);
for(kk = 1; kk < channels; kk++)
{
tx = (datadx[i*channels+kk]);
ty = (datady[i*channels+kk]);
tmp =sqrtf(tx*tx + ty*ty);
if(tmp > r[j * width + i])
{
r[j * width + i] = tmp;
c = kk;
x = tx;
y = ty;
}
}/*for(kk = 1; kk < channels; kk++)*/
max = boundary_x[0]*x + boundary_y[0]*y;
maxi = 0;
for (kk = 0; kk < CNTPARTION; kk++) {
tmp_scal = boundary_x[kk]*x + boundary_y[kk]*y;
if (tmp_scal> max) {
max = tmp_scal;
maxi = kk;
}else if (-tmp_scal> max) {
max = -tmp_scal;
maxi = kk + CNTPARTION;
}
}
alfa[j * width * 2 + i * 2 ] = maxi % CNTPARTION;
alfa[j * width * 2 + i * 2 + 1] = maxi;
}/*for(i = 0; i < width; i++)*/
}/*for(j = 0; j < height; j++)*/
//<2F><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD> <20><><EFBFBD><EFBFBD><EFBFBD> <20> <20><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
nearest_x = (int *)malloc(sizeof(int) * k);
nearest_y = (int *)malloc(sizeof(int) * k);
w = (float*)malloc(sizeof(float) * (k * 2));
for(i = 0; i < k / 2; i++)
{
nearest_x[i] = -1;
nearest_y[i] = -1;
}/*for(i = 0; i < k / 2; i++)*/
for(i = k / 2; i < k; i++)
{
nearest_x[i] = 1;
nearest_y[i] = 1;
}/*for(i = k / 2; i < k; i++)*/
for(j = 0; j < k / 2; j++)
{
b_x = k / 2 + j + 0.5f;
a_x = k / 2 - j - 0.5f;
w[j * 2 ] = 1.0f/a_x * ((a_x * b_x) / ( a_x + b_x));
w[j * 2 + 1] = 1.0f/b_x * ((a_x * b_x) / ( a_x + b_x));
}/*for(j = 0; j < k / 2; j++)*/
for(j = k / 2; j < k; j++)
{
a_x = j - k / 2 + 0.5f;
b_x =-j + k / 2 - 0.5f + k;
w[j * 2 ] = 1.0f/a_x * ((a_x * b_x) / ( a_x + b_x));
w[j * 2 + 1] = 1.0f/b_x * ((a_x * b_x) / ( a_x + b_x));
}/*for(j = k / 2; j < k; j++)*/
//<2F><><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>
for(i = 0; i < sizeY; i++)
{
for(j = 0; j < sizeX; j++)
{
for(ii = 0; ii < k; ii++)
{
for(jj = 0; jj < k; jj++)
{
if ((i * k + ii > 0) && (i * k + ii < height - 1) && (j * k + jj > 0) && (j * k + jj < width - 1))
{
d = (k*i + ii)* width + (j*k + jj);
(*map)->Map[(i ) * strsz + (j ) * (*map)->p + alfa[d * 2 ] ] +=
r[d] * w[ii * 2 ] * w[jj * 2 ];
(*map)->Map[(i ) * strsz + (j ) * (*map)->p + alfa[d * 2 + 1] + CNTPARTION] +=
r[d] * w[ii * 2 ] * w[jj * 2 ];
if ((i + nearest_y[ii] >= 0) && (i + nearest_y[ii] <= sizeY - 1))
{
(*map)->Map[(i + nearest_y[ii]) * strsz + (j ) * (*map)->p + alfa[d * 2 ] ] +=
r[d] * w[ii * 2 + 1] * w[jj * 2 ];
(*map)->Map[(i + nearest_y[ii]) * strsz + (j ) * (*map)->p + alfa[d * 2 + 1] + CNTPARTION] +=
r[d] * w[ii * 2 + 1] * w[jj * 2 ];
}
if ((j + nearest_x[jj] >= 0) && (j + nearest_x[jj] <= sizeX - 1))
{
(*map)->Map[(i ) * strsz + (j + nearest_x[jj]) * (*map)->p + alfa[d * 2 ] ] +=
r[d] * w[ii * 2 ] * w[jj * 2 + 1];
(*map)->Map[(i ) * strsz + (j + nearest_x[jj]) * (*map)->p + alfa[d * 2 + 1] + CNTPARTION] +=
r[d] * w[ii * 2 ] * w[jj * 2 + 1];
}
if ((i + nearest_y[ii] >= 0) && (i + nearest_y[ii] <= sizeY - 1) && (j + nearest_x[jj] >= 0) && (j + nearest_x[jj] <= sizeX - 1))
{
(*map)->Map[(i + nearest_y[ii]) * strsz + (j + nearest_x[jj]) * (*map)->p + alfa[d * 2 ] ] +=
r[d] * w[ii * 2 + 1] * w[jj * 2 + 1];
(*map)->Map[(i + nearest_y[ii]) * strsz + (j + nearest_x[jj]) * (*map)->p + alfa[d * 2 + 1] + CNTPARTION] +=
r[d] * w[ii * 2 + 1] * w[jj * 2 + 1];
}
}
}/*for(jj = 0; jj < k; jj++)*/
}/*for(ii = 0; ii < k; ii++)*/
}/*for(j = 1; j < sizeX - 1; j++)*/
}/*for(i = 1; i < sizeY - 1; i++)*/
cvReleaseImage(&dx);
cvReleaseImage(&dy);
free(w);
free(nearest_x);
free(nearest_y);
free(r);
free(alfa);
return LATENT_SVM_OK;
}
/*
// Feature map Normalization and Truncation
//
// API
// int normalizationAndTruncationFeatureMaps(featureMap *map, const float alfa);
// INPUT
// map - feature map
// alfa - truncation threshold
// OUTPUT
// map - truncated and normalized feature map
// RESULT
// Error status
*/
int normalizationAndTruncationFeatureMaps(CvLSVMFeatureMap *map, const float alfa)
{
int i,j, ii;
int sizeX, sizeY, p, pos, pp, xp, pos1, pos2;
float * part_noma; // norm of C(i, j)
float * new_data;
float norm_val;
sizeX = map->sizeX;
sizeY = map->sizeY;
part_noma = (float *)malloc (sizeof(float) * (sizeX * sizeY));
p = map->xp / 3;
for(i = 0; i < sizeX * sizeY; i++)
{
norm_val = 0.0;
pos = i * map->p;
for(j = 0; j < p; j++)
{
norm_val += map->Map[pos + j] * map->Map[pos + j];
}/*for(j = 0; j < p; j++)*/
part_noma[i] = norm_val;
}/*for(i = 0; i < sizeX * sizeY; i++)*/
xp = map->xp;
pp = xp * 4;
sizeX -= 2;
sizeY -= 2;
new_data = (float *)malloc (sizeof(float) * (sizeX * sizeY * pp));
//normalization
for(i = 1; i <= sizeY; i++)
{
for(j = 1; j <= sizeX; j++)
{
norm_val = sqrtf(
part_noma[(i )*(sizeX + 2) + (j )] +
part_noma[(i )*(sizeX + 2) + (j + 1)] +
part_noma[(i + 1)*(sizeX + 2) + (j )] +
part_noma[(i + 1)*(sizeX + 2) + (j + 1)]);
pos1 = (i ) * (sizeX + 2) * xp + (j ) * xp;
pos2 = (i-1) * (sizeX ) * pp + (j-1) * pp;
for(ii = 0; ii < p; ii++)
{
new_data[pos2 + ii ] = map->Map[pos1 + ii ] / norm_val;
}/*for(ii = 0; ii < p; ii++)*/
for(ii = 0; ii < 2 * p; ii++)
{
new_data[pos2 + ii + p * 4] = map->Map[pos1 + ii + p] / norm_val;
}/*for(ii = 0; ii < 2 * p; ii++)*/
norm_val = sqrtf(
part_noma[(i )*(sizeX + 2) + (j )] +
part_noma[(i )*(sizeX + 2) + (j + 1)] +
part_noma[(i - 1)*(sizeX + 2) + (j )] +
part_noma[(i - 1)*(sizeX + 2) + (j + 1)]);
for(ii = 0; ii < p; ii++)
{
new_data[pos2 + ii + p ] = map->Map[pos1 + ii ] / norm_val;
}/*for(ii = 0; ii < p; ii++)*/
for(ii = 0; ii < 2 * p; ii++)
{
new_data[pos2 + ii + p * 6] = map->Map[pos1 + ii + p] / norm_val;
}/*for(ii = 0; ii < 2 * p; ii++)*/
norm_val = sqrtf(
part_noma[(i )*(sizeX + 2) + (j )] +
part_noma[(i )*(sizeX + 2) + (j - 1)] +
part_noma[(i + 1)*(sizeX + 2) + (j )] +
part_noma[(i + 1)*(sizeX + 2) + (j - 1)]);
for(ii = 0; ii < p; ii++)
{
new_data[pos2 + ii + p * 2] = map->Map[pos1 + ii ] / norm_val;
}/*for(ii = 0; ii < p; ii++)*/
for(ii = 0; ii < 2 * p; ii++)
{
new_data[pos2 + ii + p * 8] = map->Map[pos1 + ii + p] / norm_val;
}/*for(ii = 0; ii < 2 * p; ii++)*/
norm_val = sqrtf(
part_noma[(i )*(sizeX + 2) + (j )] +
part_noma[(i )*(sizeX + 2) + (j - 1)] +
part_noma[(i - 1)*(sizeX + 2) + (j )] +
part_noma[(i - 1)*(sizeX + 2) + (j - 1)]);
for(ii = 0; ii < p; ii++)
{
new_data[pos2 + ii + p * 3 ] = map->Map[pos1 + ii ] / norm_val;
}/*for(ii = 0; ii < p; ii++)*/
for(ii = 0; ii < 2 * p; ii++)
{
new_data[pos2 + ii + p * 10] = map->Map[pos1 + ii + p] / norm_val;
}/*for(ii = 0; ii < 2 * p; ii++)*/
}/*for(j = 1; j <= sizeX; j++)*/
}/*for(i = 1; i <= sizeY; i++)*/
//truncation
for(i = 0; i < sizeX * sizeY * pp; i++)
{
if(new_data [i] > alfa) new_data [i] = alfa;
}/*for(i = 0; i < sizeX * sizeY * pp; i++)*/
//swop data
map->p = pp;
map->xp = xp;
map->sizeX = sizeX;
map->sizeY = sizeY;
free (map->Map);
free (part_noma);
map->Map = new_data;
return LATENT_SVM_OK;
}
/*
// Feature map reduction
// In each cell we reduce dimension of the feature vector
// according to original paper special procedure
//
// API
// int PCAFeatureMaps(featureMap *map)
// INPUT
// map - feature map
// OUTPUT
// map - feature map
// RESULT
// Error status
*/
int PCAFeatureMaps(CvLSVMFeatureMap *map)
{
int i,j, ii, jj, k;
int sizeX, sizeY, p, pp, xp, yp, pos1, pos2;
float * new_data;
float val;
float nx, ny;
sizeX = map->sizeX;
sizeY = map->sizeY;
p = map->p;
pp = map->xp + 4;
yp = 4;
xp = (map->xp / 3);
nx = 1.0f / sqrtf((float)(xp * 2));
ny = 1.0f / sqrtf((float)(yp ));
new_data = (float *)malloc (sizeof(float) * (sizeX * sizeY * pp));
for(i = 0; i < sizeY; i++)
{
for(j = 0; j < sizeX; j++)
{
pos1 = ((i)*sizeX + j)*p;
pos2 = ((i)*sizeX + j)*pp;
k = 0;
for(jj = 0; jj < xp * 2; jj++)
{
val = 0;
for(ii = 0; ii < yp; ii++)
{
val += map->Map[pos1 + yp * xp + ii * xp * 2 + jj];
}/*for(ii = 0; ii < yp; ii++)*/
new_data[pos2 + k] = val * ny;
k++;
}/*for(jj = 0; jj < xp * 2; jj++)*/
for(jj = 0; jj < xp; jj++)
{
val = 0;
for(ii = 0; ii < yp; ii++)
{
val += map->Map[pos1 + ii * xp + jj];
}/*for(ii = 0; ii < yp; ii++)*/
new_data[pos2 + k] = val * ny;
k++;
}/*for(jj = 0; jj < xp; jj++)*/
for(ii = 0; ii < yp; ii++)
{
val = 0;
for(jj = 0; jj < 2 * xp; jj++)
{
val += map->Map[pos1 + yp * xp + ii * xp * 2 + jj];
}/*for(jj = 0; jj < xp; jj++)*/
new_data[pos2 + k] = val * nx;
k++;
} /*for(ii = 0; ii < yp; ii++)*/
}/*for(j = 0; j < sizeX; j++)*/
}/*for(i = 0; i < sizeY; i++)*/
//swop data
map->p = pp;
map->xp = pp;
free (map->Map);
map->Map = new_data;
return LATENT_SVM_OK;
}
/*
// Getting feature pyramid
//
// API
// int getFeaturePyramid(IplImage * image, const filterObject **all_F,
const int n_f,
const int lambda, const int k,
const int startX, const int startY,
const int W, const int H, featurePyramid **maps);
// INPUT
// image - image
// lambda - resize scale
// k - size of cells
// startX - X coordinate of the image rectangle to search
// startY - Y coordinate of the image rectangle to search
// W - width of the image rectangle to search
// H - height of the image rectangle to search
// OUTPUT
// maps - feature maps for all levels
// RESULT
// Error status
*/
int getFeaturePyramid(IplImage * image,
const int lambda, const int k,
const int startX, const int startY,
const int W, const int H, CvLSVMFeaturePyramid **maps)
{
IplImage *img2, *imgTmp, *imgResize;
float step, tmp;
int cntStep;
int maxcall;
int i;
int err;
CvLSVMFeatureMap *map;
//geting subimage
cvSetImageROI(image, cvRect(startX, startY, W, H));
img2 = cvCreateImage(cvGetSize(image), image->depth, image->nChannels);
cvCopy(image, img2, NULL);
cvResetImageROI(image);
if(img2->depth != IPL_DEPTH_32F)
{
imgResize = cvCreateImage(cvSize(img2->width , img2->height) , IPL_DEPTH_32F , 3);
cvConvert(img2, imgResize);
}
else
{
imgResize = img2;
}
step = powf(2.0f, 1.0f/ ((float)lambda));
maxcall = W/k;
if( maxcall > H/k )
{
maxcall = H/k;
}
cntStep = (int)(logf((float)maxcall/(5.0f))/logf(step)) + 1;
//printf("Count step: %f %d\n", step, cntStep);
allocFeaturePyramidObject(maps, lambda, cntStep + lambda);
for(i = 0; i < lambda; i++)
{
tmp = 1.0f / powf(step, (float)i);
imgTmp = resize_opencv (imgResize, tmp);
//imgTmp = resize_article_dp(img2, tmp, 4);
err = getFeatureMaps_dp(imgTmp, 4, &map);
err = normalizationAndTruncationFeatureMaps(map, 0.2f);
err = PCAFeatureMaps(map);
(*maps)->pyramid[i] = map;
//printf("%d, %d\n", map->sizeY, map->sizeX);
cvReleaseImage(&imgTmp);
}
/**********************************one**************/
for(i = 0; i < cntStep; i++)
{
tmp = 1.0f / powf(step, (float)i);
imgTmp = resize_opencv (imgResize, tmp);
//imgTmp = resize_article_dp(imgResize, tmp, 8);
err = getFeatureMaps_dp(imgTmp, 8, &map);
err = normalizationAndTruncationFeatureMaps(map, 0.2f);
err = PCAFeatureMaps(map);
(*maps)->pyramid[i + lambda] = map;
//printf("%d, %d\n", map->sizeY, map->sizeX);
cvReleaseImage(&imgTmp);
}/*for(i = 0; i < cntStep; i++)*/
if(img2->depth != IPL_DEPTH_32F)
{
cvReleaseImage(&imgResize);
}
cvReleaseImage(&img2);
return LATENT_SVM_OK;
}
/*
// add zero border to feature map
//
// API
// int addBordersToFeatureMaps(featureMap *map, const int bX, const int bY);
// INPUT
// map - feature map
// bX - border size in x
// bY - border size in y
// OUTPUT
// map - feature map
// RESULT
// Error status
*/
int addBordersToFeatureMaps(CvLSVMFeatureMap *map, const int bX, const int bY){
int i,j, jj;
int sizeX, sizeY, p, pos1, pos2;
float * new_data;
sizeX = map->sizeX;
sizeY = map->sizeY;
p = map->p;
new_data = (float *)malloc (sizeof(float) * ((sizeX + 2 * bX) * (sizeY + 2 * bY) * p));
for(i = 0; i < ((sizeX + 2 * bX) * (sizeY + 2 * bY) * p); i++)
{
new_data[i] = (float)0;
}/*for(i = 0; i < ((sizeX + 2 * bX) * (sizeY + 2 * bY) * p); i++)*/
for(i = 0; i < sizeY; i++)
{
for(j = 0; j < sizeX; j++)
{
pos1 = ((i )*sizeX + (j )) * p;
pos2 = ((i + bY)*(sizeX + 2 * bX) + (j + bX)) * p;
for(jj = 0; jj < p; jj++)
{
new_data[pos2 + jj] = map->Map[pos1 + jj];
}/*for(jj = 0; jj < p; jj++)*/
}/*for(j = 0; j < sizeX; j++)*/
}/*for(i = 0; i < sizeY; i++)*/
//swop data
map->sizeX = sizeX + 2 * bX;
map->sizeY = sizeY + 2 * bY;
free (map->Map);
map->Map = new_data;
return LATENT_SVM_OK;
2010-10-21 20:30:57 +02:00
}