/*
This function takes in a window(64x128 pixels,
but can be easily modified for other window sizes)
and calculated the hog features for the window. It
can be used to calculate the feature vector for a
64x128 pixel image as well. This window/image is the
training/detection window which is used for training
or on which the final detection is done. The hog features
are computed by dividing the window into overlapping blocks,
calculating the hog vectors for each block using calculateHOG_block(...)
and concatenating the so obtained vectors to obtain the
hog feature vector for the window
*/
CvMat* HoGProcessor::calculateHOG_window(IplImage **integrals, CvRect window, int normalization){
/*
A cell size of 8x8 pixels is considered and each block
is divided into 2x2 such cells(i.e. the block is 16x16 pixels).
So a 64x128 pixels window would be divided into 7x15 overlapping blocks
*/
int block_start_x, block_start_y;
/* The length of the feature vector for a cell is 9
(since no. of bins is 9), for block it would be
9*(no. of cells in the block) = 9*4 = 36. And the lenght
of the feature vector for a window would be 36*(no. of
blocks in the window)
*/
int lenghth_feature_of_block = 9 * m_block.height * m_block.width;
int num_overlap_block_of_width = ((window.width - m_cell.width * m_block.width) / (m_cell.width * m_fStepOverlap)) + 1;
int num_overlap_block_of_height = ((window.height - m_cell.height * m_block.height)/ (m_cell.height * m_fStepOverlap)) + 1;
int total_block_of_window = num_overlap_block_of_width * num_overlap_block_of_height ;
CvMat* window_feature_vector = cvCreateMat(1
, total_block_of_window * lenghth_feature_of_block
, CV_32FC1);
CvMat vector_block;
int startcol = 0;
int widthBlockInPixels = m_cell.width * m_block.width;
int heightBlockInPixels = m_cell.height * m_block.height;
for(block_start_y = window.y;
block_start_y <= window.y + window.height - m_cell.height * m_block.height;
block_start_y += m_cell.height * m_fStepOverlap)
{//overllap
for(block_start_x = window.x;
block_start_x <= window.x + window.width - m_cell.width * m_block.width;
block_start_x += m_cell.width * m_fStepOverlap)
{//overllap
cvGetCols(window_feature_vector, &vector_block, startcol, startcol+36);
calculateHOG_block(
cvRect(block_start_x, block_start_y, widthBlockInPixels , heightBlockInPixels)
, &vector_block, integrals, m_cell
, normalization);
startcol+= lenghth_feature_of_block;
}
}
return window_feature_vector;
}
CvMat* calculateHOG_window(IplImage** integrals,
CvRect window, int normalization)
{
/*A cell size of 8x8 pixels is considered and each
block is divided into 2x2 such cells (i.e. the block
is 16x16 pixels). So a 64x128 pixels window would be
divided into 7x15 overlapping blocks*/
int block_start_x, block_start_y, cell_width = 8;
int cell_height = 8;
int block_width = 2, block_height = 2;
/* The length of the feature vector for a cell is
9(since no. of bins is 9), for block it would be
9*(no. of cells in the block) = 9*4 = 36. And the
length of the feature vector for a window would be
36*(no. of blocks in the window */
CvMat* window_feature_vector = cvCreateMat(1,
((((window.width - cell_width * block_width)
/ cell_width) + 1) * (((window.height -
cell_height * block_height) / cell_height)
+ 1)) * 36, CV_32FC1);
CvMat vector_block;
int startcol = 0;
for (block_start_y = window.y; block_start_y
<= window.y + window.height - cell_height
* block_height; block_start_y += cell_height)
{
for (block_start_x = window.x; block_start_x
<= window.x + window.width - cell_width
* block_width; block_start_x += cell_width)
{
cvGetCols(window_feature_vector, &vector_block,
startcol, startcol + 36);
calculateHOG_block(cvRect(block_start_x,
block_start_y, cell_width * block_width, cell_height
* block_height), &vector_block, integrals, cvSize(
cell_width, cell_height), normalization);
startcol += 36;
}
}
return (window_feature_vector);
}
CvMat* HoGProcessor::calculateHOG_window(IplImage *img, IplImage **integrals, CvRect window, int normalization){
/*
A cell size of 8x8 pixels is considered and each block
is divided into 2x2 such cells(i.e. the block is 16x16 pixels).
So a 64x128 pixels window would be divided into 7x15 overlapping blocks
*/
int block_start_x, block_start_y, cell_width = m_cell.width;
int cell_height = m_cell.height;
int block_width = m_block.width, block_height = m_block.height;
/* The length of the feature vector for a cell is 9
(since no. of bins is 9), for block it would be
9*(no. of cells in the block) = 9*4 = 36. And the lenght
of the feature vector for a window would be 36*(no. of
blocks in the window)
*/
int lenghth_feature_of_block = 9 * m_block.height * m_block.width;
int num_overlap_block_of_width = ((window.width - cell_width * block_width) / (cell_width * m_fStepOverlap)) + 1;
int num_overlap_block_of_height = ((window.height - cell_height * block_height)/ (cell_height * m_fStepOverlap)) + 1;
int total_block_of_window = num_overlap_block_of_width * num_overlap_block_of_height ;
CvMat* window_feature_vector = cvCreateMat(1
, total_block_of_window * lenghth_feature_of_block
, CV_32FC1);
CvMat vector_block;
int startcol = 0;
cvNamedWindow("aaa");
IplImage *imgHog = cvCreateImage(cvSize(9,9), img->depth, 1);
IplImage *imgHogShow = cvCreateImage(cvSize(81,81), img->depth, 1);
for(block_start_y = window.y;
block_start_y <= window.y + window.height - cell_height * block_height;
block_start_y += cell_height * m_fStepOverlap)
{//overllap
for(block_start_x = window.x;
block_start_x <= window.x + window.width - cell_width * block_width;
block_start_x += cell_width * m_fStepOverlap)
{//overllap
cvGetCols(window_feature_vector, &vector_block, startcol, startcol+36);
calculateHOG_block(cvRect(block_start_x,
block_start_y, cell_width * block_width, cell_height
* block_height), &vector_block, integrals, cvSize(
cell_width, cell_height), normalization);
startcol += lenghth_feature_of_block;
int row=0;
int col=0;
for(row = 0; row < 9; row++)
{
for(col = 0; col < 9; col++)
{
float cell1 = CV_MAT_ELEM(vector_block, float, 0, (9 * row + col));
CV_IMAGE_ELEM(imgHog, uchar, row, col) = cell1 * 255;
}
}
cvResize(imgHog, imgHogShow);
cvShowImage("aaa", imgHogShow);
cvWaitKey();
}
}
cvReleaseImage(&imgHog);
cvReleaseImage(&imgHogShow);
return window_feature_vector;
}
