shared_ptr<Image3>& ImageStats::getImageVar(int ft, bool normalize){
if (normalize){
for (int c = 0 ; c < COLOR_CHS ; c++){
m_varMin[ft][c] = std::numeric_limits<float>::max();
m_varMax[ft][c] = std::numeric_limits<float>::min();
for (int x = 0 ; x < m_w ; x++){
for (int y = 0 ; y < m_h ; y++){
if ( getVariance(x,y,ft,c) < m_varMin[ft][c]) m_varMin[ft][c] = getVariance(x,y,ft,c);
if ( getVariance(x,y,ft,c) > m_varMax[ft][c]) m_varMax[ft][c] = getVariance(x,y,ft,c);
}
}
}
for (int x = 0 ; x < m_w ; x++)
for (int y = 0 ; y < m_h ; y++)
m_imageVar[ft]->fastSet(x,y,Color3((getVariance(x,y,ft,RED) - m_varMin[ft][RED])/(m_varMax[ft][RED] - m_varMin[ft][RED]),
(getVariance(x,y,ft,GREEN) - m_varMin[ft][GREEN])/(m_varMax[ft][GREEN] - m_varMin[ft][GREEN]),
(getVariance(x,y,ft,BLUE) - m_varMin[ft][BLUE])/(m_varMax[ft][BLUE] - m_varMin[ft][BLUE])));
}
else{
for (int x = 0 ; x < m_w ; x++)
for (int y = 0 ; y < m_h ; y++)
m_imageVar[ft]->fastSet(x,y,Color3(getVariance(x,y,ft,RED),getVariance(x,y,ft,GREEN),getVariance(x,y,ft,BLUE)));
}
return m_imageVar[ft];
}
double
StatisticalTimer::getStdDev( sTimerID id ) const
{
double variance = getVariance( id );
return sqrt( variance );
}
double
statTimer::getStdDev( size_t id ) const
{
double variance = getVariance( id );
return sqrt( variance );
}
float FloatArray::getStandardDeviation(){
float result;
/// @note When built for ARM Cortex-M processor series, this method uses the optimized <a href="http://www.keil.com/pack/doc/CMSIS/General/html/index.html">CMSIS library</a>
#ifdef ARM_CORTEX
arm_std_f32 (data, size, &result);
#else
result=sqrtf(getVariance());
#endif
return result;
}
cv::Rect selectBestRectByVariance(std::vector<cv::Point>& poly, std::vector<cv::Rect>& vrect, cv::Rect cur)
{
double min=-1;
int idx=-1;
for(int i=0; i<vrect.size(); i++){
if(isRectInPolygon(poly, vrect[i])){
if(min<0){
min=getVariance(poly, vrect[i]);
idx=i;
}else{
double var=getVariance(poly, vrect[i]);
if(var<min){ min=var; idx=i; }
}
}
}
if(idx>=0) return vrect[idx];
else return cur;
}
bool RollingValueStore :: getStdDev(char* key,float* result)
{
bool isVarianceSuccessfullyRetrieved = getVariance(key,result);
if(isVarianceSuccessfullyRetrieved) {
*result = sqrt(*result);
return true;
}
else {
return false;
}
}
void Statistics::scaling(vector<double>& v)
{
double std = sqrt(getVariance(v));
// standard deviation = 0, i.e. all values of this vector are identical, so do nothing!
if (std < 5*std::numeric_limits < double > ::epsilon()) return;
for (unsigned int i = 0; i < v.size(); i++)
{
v[i] /= std;
}
}
void CybMLM::checkVariables()
{
for(int i=0; i < getVariablesNumber(); i++)
if(getVariance(i))
this->variables->insert(i);
this->precisionMatrix = new float[getVariablesNumber()*getVariablesNumber()];
this->determinat = CybMatrixOperator::matrixDeterminant(covariance, variables->size());
CybMatrixOperator::matrixInverse(covariance,this->precisionMatrix,variables->size(),determinat);
}
std::vector< StatData >
GpuStatTimer::getStdDev( size_t id )
{
std::vector< StatData > stddev = getVariance( id );
for( cl_uint v = 0; v < stddev.size( ); ++v )
{
stddev[ v ].doubleNanoSec = sqrt( stddev[ v ].doubleNanoSec );
}
return stddev;
}
void inline unitTestKMeansRandomly(Group &test, std::ofstream &output) {
// 数据归一化
// normaliztion(test);
// 输出读入的数据
// puts("Print Input Data:");
// puts("=====================================");
// test.display();
// puts("=====================================\n");
// k 值
const unsigned k = 3;
/** 准备测试数据 */
// KMeans++
// std::vector<Group> centroid = buildInitialPointRandomly(k, test);
// Kmeans + Density initialize
// std::vector<Group> centroid = buildInitialPointDensity(k, test);
// KMeans
// std::vector<Group> centroid = buildInitialPoint(k, test);
output << "===============================================" << std::endl;
output << "buildInitialPointRandomly" << std::endl;
output << "===============================================" << std::endl;
// 重复运行测试
std::vector<Group> result;
std::vector<double> avg(k);
double SSE = 0;
// 记录每次的 SSE 值
std::vector<double> vecSSE;
time_t start = std::clock();
// 重复实验的次数
for (unsigned i = 0; i < TIMES; ++i) {
printf("running: %d\r", i);
result = KMeans(test, k, buildInitialPointRandomly(k, test), false);
SSE = 0;
for (unsigned j = 0; j < result.size(); ++j) {
SSE += result[j].getSumOfEuclideanDistance();
vecSSE.push_back(SSE / result.size());
}
}
time_t end = std::clock();
output << "The average of SSE = " << getAverage(vecSSE) << std::endl;
output << "The variance of SSE = " << getVariance(vecSSE) << std::endl;
output << "The running time is: " << double(end - start) / CLOCKS_PER_SEC << std::endl;
// printf("The average of SSE = %lf\n", getAverage(vecSSE));
// printf("The variance of SSE = %lf\n", getVariance(vecSSE));
printf("the running time is : %f\n", double(end - start) / CLOCKS_PER_SEC);
output << "===============================================" << std::endl;
// puts("===============================================");
// KMedoids(test, k, centroid);
}
void ParticleManager::setVisibleNParticles(int n)
{
int activatedParticles = 0;
for (int i = 0; i < MAX_NUMBER_PARTICLES; i++)
{
if (!particles[i].getActive()) //found an inactive particle
{
particles[i].setActive(true);
particles[i].setMaxTimeAlive(MAX_PARTICLE_LIFETIME*getVariance());
float newX = velocity.x * getVariance();
float newY = velocity.y * getVariance();
VECTOR2 v = VECTOR2(newX,newY);
particles[i].setX(position.x);
particles[i].setY(position.y);
particles[i].setVelocity(v);
particles[i].setVisible(true);
activatedParticles++;
if (activatedParticles == n)
return;
}
}
}
double BaseStat::getConfInterval(CONFIDENCE_INTERVAL c)
{
double mu; // the mean
double s; // the variance
if (!_endOfSim) throw Exc(GET);
if (!_initFlag) throw Exc(NO_INIT);
if (_expNum < 3) throw Exc(NEED_3);
mu = getMean();
s = getVariance();
return t_student(c, (unsigned int) _expNum - 1) * s;
}
void Statistics::centering(Eigen::MatrixXd& m, int col)
{
double mean = getMean(m, col);
double std = sqrt(getVariance(m, col, mean));
// standard deviation = 0, i.e. all values of this column are identical, so do nothing!
if (std < 5*std::numeric_limits < double > ::epsilon()) return;
for (int i = 0; i < m.rows(); i++)
{
m(i, col) = (m(i, col)-mean)/std;
}
}
void Statistics::centering(vector<double>& v, double& mean, double& std)
{
mean = getMean(v);
std = sqrt(getVariance(v, mean));
// standard deviation = 0, i.e. all values of this vector are identical, so do nothing!
if (std < 5*std::numeric_limits < double > ::epsilon()) return;
for (unsigned int i = 0; i < v.size(); i++)
{
v[i] = (v[i]-mean)/std;
}
}
PRFloat64 getStdDev(VarianceState* inVariance)
/*
** Determine the standard deviation based on the given state.
*/
{
PRFloat64 retval = 0.0;
PRFloat64 variance;
variance = getVariance(inVariance);
retval = sqrt(variance);
return retval;
}
std::unique_ptr<Image> intensityPatch(const std::unique_ptr<Image>& image,
float mean, float stdev)
{
if (image->getChannelNum() != 1)
return std::unique_ptr<Image>();
const auto orig_data_ptr = image->getValues(0);
const auto w = image->getWidth();
const auto h = image->getHeight();
float current_mean = 0;
for (size_t j = 0; j < h; j++)
{
for (size_t i = 0; i < w; i++)
{
current_mean += orig_data_ptr[j*w + i];
}
}
current_mean /= (w*h);
float current_var = getVariance(image);
if (std::isfinite(current_var) && !std::isnormal(current_var))
{
// if the sum of squares of errors is equal to zero or is subnormal,
// then just fill the result image with (0.5) in order to aviod
// divide-by-zero exception
auto newImage = std::make_unique<Image>(w, h, 1);
for (size_t j = 0; j < h; j++)
{
for (size_t i = 0; i < w; i++)
{
(newImage->getValues(0))[j*w + i] = 0.5;
}
}
return newImage;
}
float alpha = stdev * std::sqrt(1.0 / current_var);
auto newImage = std::make_unique<Image>(w, h, 1);
for (size_t j = 0; j < h; j++)
{
for (size_t i = 0; i < w; i++)
{
(newImage->getValues(0))[j*w + i] = alpha * (orig_data_ptr[j*w + i]
- current_mean) + mean;
}
}
return newImage;
}
double GROUP::calculateUCB(int total)
{
//ucb_tuned1
double v = getVariance() + sqrt(2*log(total) / AllUsedNumber);
double V = (v>0.25) ? 0.25 : v;
double max = getmax() , min = getmin();
double newMean;
if(max == min)
newMean = 1.0;
else
newMean = (getMean() - min ) / (max - min);
double tmp = newMean - sqrt(log(total) / AllUsedNumber *V);
return tmp;
}
virtual int selectNextArm(){
double n = vectorSum(Ni);
std::vector<double> indices = std::vector<double>(K, 0.0);
for(uint k=0;k<K;++k){
if(Ni[k]==0){
return k;
}
double variance = getVariance(k);
//UCB-V index
indices[k] = (Gi[k]/Ni[k]) + sqrt( (2 * zeta * variance * log(n)) / (double)Ni[k] ) + 3 * amp * zeta * log(n)/(double)Ni[k];
}
int targetArm = vectorMaxIndex(indices);
return targetArm;
}
PriorBoxLayerImpl(const LayerParams ¶ms)
{
setParamsFrom(params);
_minSize = getParameter<unsigned>(params, "min_size");
CV_Assert(_minSize > 0);
_flip = getParameter<bool>(params, "flip");
_clip = getParameter<bool>(params, "clip");
_aspectRatios.clear();
_aspectRatios.push_back(1.);
getAspectRatios(params);
getVariance(params);
_numPriors = _aspectRatios.size();
_maxSize = -1;
if (params.has("max_size"))
{
_maxSize = params.get("max_size").get<float>(0);
CV_Assert(_maxSize > _minSize);
_numPriors += 1;
}
if (params.has("step_h") || params.has("step_w")) {
CV_Assert(!params.has("step"));
_stepY = getParameter<float>(params, "step_h");
CV_Assert(_stepY > 0.);
_stepX = getParameter<float>(params, "step_w");
CV_Assert(_stepX > 0.);
} else if (params.has("step")) {
const float step = getParameter<float>(params, "step");
CV_Assert(step > 0);
_stepY = step;
_stepX = step;
} else {
_stepY = 0;
_stepX = 0;
}
}
PriorBoxLayerImpl(const LayerParams ¶ms)
: _boxWidth(0), _boxHeight(0)
{
setParamsFrom(params);
_minSize = getParameter<float>(params, "min_size", 0, false, 0);
_flip = getParameter<bool>(params, "flip", 0, false, true);
_clip = getParameter<bool>(params, "clip", 0, false, true);
_bboxesNormalized = getParameter<bool>(params, "normalized_bbox", 0, false, true);
_scales.clear();
_aspectRatios.clear();
getAspectRatios(params);
getVariance(params);
getParams("scales", params, &_scales);
getParams("width", params, &_widths);
getParams("height", params, &_heights);
_explicitSizes = !_widths.empty();
CV_Assert(_widths.size() == _heights.size());
if (_explicitSizes)
{
CV_Assert(_aspectRatios.empty(), !params.has("min_size"), !params.has("max_size"));
_numPriors = _widths.size();
}
else
{
CV_Assert(!_aspectRatios.empty(), _minSize > 0);
_numPriors = _aspectRatios.size() + 1; // + 1 for an aspect ratio 1.0
}
_maxSize = -1;
if (params.has("max_size"))
{
_maxSize = params.get("max_size").get<float>(0);
CV_Assert(_maxSize > _minSize);
_numPriors += 1;
}
if (params.has("step_h") || params.has("step_w")) {
CV_Assert(!params.has("step"));
_stepY = getParameter<float>(params, "step_h");
CV_Assert(_stepY > 0.);
_stepX = getParameter<float>(params, "step_w");
CV_Assert(_stepX > 0.);
} else if (params.has("step")) {
const float step = getParameter<float>(params, "step");
CV_Assert(step > 0);
_stepY = step;
_stepX = step;
} else {
_stepY = 0;
_stepX = 0;
}
if (params.has("offset_h") || params.has("offset_w"))
{
CV_Assert(!params.has("offset"), params.has("offset_h"), params.has("offset_w"));
getParams("offset_h", params, &_offsetsY);
getParams("offset_w", params, &_offsetsX);
CV_Assert(_offsetsX.size() == _offsetsY.size());
_numPriors *= std::max((size_t)1, 2 * (_offsetsX.size() - 1));
}
else
{
float offset = getParameter<float>(params, "offset", 0, false, 0.5);
_offsetsX.assign(1, offset);
_offsetsY.assign(1, offset);
}
}
double Variable::getStddev() const
{
return std::sqrt(getVariance());
}
void Accumulator::getStdDev(double &val) const
{
getVariance(val);
val = sqrt(val);
}
void makeTree(node* root, int* aSet, int rows, int cols)
{
int* vector = NULL;
double* varianceVector;
double* varianceVectorCpy;
int* maxVarianceVector;
double maxVariance;
int maxVarianceIdx;
double* median;
int i;
int leftSize = 0;
int rightSize = 0;
double vectorVal;
int* leftSet;
int* rightSet;
node* leftNode;
node* rightNode;
int j = 0;
int a = 0;
leftNode = (node*) malloc(sizeof(node));
leftNode->dimNumber = -1;
leftNode->dimVal = -1;
leftNode->left = NULL;
leftNode->right = NULL;
leftNode->isLeaf = 0;
leftNode->leafVector = NULL;
leftNode->id = NULL;
rightNode = (node*) malloc(sizeof(node));
rightNode->dimNumber = -1;
rightNode->dimVal = -1;
rightNode->left = NULL;
rightNode->right = NULL;
rightNode->isLeaf = 0;
rightNode->leafVector = NULL;
rightNode->id = NULL;
//printf("nodes left: %d\n", nodes);
//vector = getDimensionVector(1, aSet, rows, cols);
varianceVector = (double*) malloc(cols * sizeof(double) );
median = (double*) malloc(sizeof(double));
if (median == NULL)
{
a = 1;
}
for (i = 0; i < cols; i++)
{
vector = getDimensionVector(i, aSet, rows, cols);
getVariance(&varianceVector[i], vector, rows);
free(vector);
}
varianceVectorCpy = (double*) malloc(cols*sizeof(double));
varianceVectorCpy = (double*) memcpy(varianceVectorCpy, varianceVector, cols*sizeof(double));
qsort(varianceVectorCpy, cols, sizeof(double), dCmpfunc);
//for (i = 0; i < cols; i++)
//{
// printf("%f ", varianceVectorCpy[i]);
//}
for (i = 0; i < cols; i++)
{
maxVariance = varianceVectorCpy[cols-1-i];
if (maxVariance != 0.0)
{
break;
}
}
//if (maxVariance == 0.0)
//{
// for (i = 0; i < cols; i++)
// {
// printf("%f ", varianceVectorCpy[i]);
// }
// printf("\n");
// printf("\n");
// printf("------------------------");
// for (i = 0; i < cols; i++)
// {
// printf("%f ", varianceVector[i]);
// }
// printf("\n");
// printf("\n");
// printf("\n");
// for (i =0; i < rows; i++)
// {
// for (j = 0; j < cols; j++)
// {
// printf("%d ", aSet[i+j]);
// }
// printf("\n");
// }
// a = 1;
//}
// find dimension number with max variance
for (i = 0; i < cols; i++)
{
if (varianceVector[i] == maxVariance )
{
maxVarianceIdx = i;
}
}
// get max variance axe
maxVarianceVector = getDimensionVector(maxVarianceIdx, aSet, rows, cols);
getMedian(median, maxVarianceVector, rows);
// create two arrays
// determine array sizes
for (i = 0; i < rows; i++)
{
vectorVal = (double) maxVarianceVector[i];
if (vectorVal <= *median)
{ leftSize++; }
else
{ rightSize++; }
}
// set root values
root->dimNumber = maxVarianceIdx;
root->dimVal = *median;
// get memory from branches
leftSet = (int*) malloc(cols * leftSize * sizeof(int));
rightSet = (int*) malloc(cols * rightSize * sizeof(int));
// reset counters
leftSize = 0;
rightSize = 0;
// split aSet into two parts according to median value
for (i = 0; i < rows; i++)
{
vectorVal = (double) maxVarianceVector[i];
if (vectorVal <= *median)
{
memcpy(&leftSet[leftSize * cols], &aSet[i * cols], cols * sizeof(int));
leftSize++;
}
else
{
memcpy(&rightSet[rightSize * cols], &aSet[i * cols], cols * sizeof(int));
rightSize++;
}
}
//free(vector);
free(varianceVector);
free(varianceVectorCpy);
free(maxVarianceVector);
free(aSet); // delete input set (matrix) to store some memory. In futher calculation it is not used. Used only its split version
free(median);
// making a leaf or node for right and left root branches
if (leftSize > 1)
{
root->left = leftNode;
makeTree(leftNode, leftSet, leftSize, cols);
}
else
{
leftNode->isLeaf = 1;
leftNode->leafVector = leftSet;
root->left = leftNode;
nodes--;
}
if (rightSize > 1)
{
root->right = rightNode;
makeTree(rightNode, rightSet, rightSize, cols );
}
else
{
rightNode->isLeaf = 1;
rightNode->leafVector = rightSet;
root->right = rightNode;
nodes--;
}
}
double CSimpleStat::getStdev() const
{
return sqrt(std::max(0.0, getVariance()));
}
double Statistics::getStddev(const vector<double>& v, double mean)
{
double var = getVariance(v, mean);
return sqrt(var);
}
double
Statistic::getStandardDeviation() const {
return sqrt(getVariance());
}
double Statistics::getStddev(const Eigen::MatrixXd& m, int col, double mean)
{
double d = getVariance(m, col, mean);
return sqrt(d);
}
double getStddev() const { return sqrt(getVariance()); }
PriorBoxLayerImpl(const LayerParams ¶ms)
{
setParamsFrom(params);
_minSize = getParameter<float>(params, "min_size", 0, false, 0);
_flip = getParameter<bool>(params, "flip", 0, false, true);
_clip = getParameter<bool>(params, "clip", 0, false, true);
_bboxesNormalized = getParameter<bool>(params, "normalized_bbox", 0, false, true);
_aspectRatios.clear();
getAspectRatios(params);
getVariance(params);
_maxSize = -1;
if (params.has("max_size"))
{
_maxSize = params.get("max_size").get<float>(0);
CV_Assert(_maxSize > _minSize);
}
std::vector<float> widths, heights;
getParams("width", params, &widths);
getParams("height", params, &heights);
_explicitSizes = !widths.empty();
CV_Assert(widths.size() == heights.size());
if (_explicitSizes)
{
CV_Assert(_aspectRatios.empty());
CV_Assert(!params.has("min_size"));
CV_Assert(!params.has("max_size"));
_boxWidths = widths;
_boxHeights = heights;
}
else
{
CV_Assert(_minSize > 0);
_boxWidths.resize(1 + (_maxSize > 0 ? 1 : 0) + _aspectRatios.size());
_boxHeights.resize(_boxWidths.size());
_boxWidths[0] = _boxHeights[0] = _minSize;
int i = 1;
if (_maxSize > 0)
{
// second prior: aspect_ratio = 1, size = sqrt(min_size * max_size)
_boxWidths[i] = _boxHeights[i] = sqrt(_minSize * _maxSize);
i += 1;
}
// rest of priors
for (size_t r = 0; r < _aspectRatios.size(); ++r)
{
float arSqrt = sqrt(_aspectRatios[r]);
_boxWidths[i + r] = _minSize * arSqrt;
_boxHeights[i + r] = _minSize / arSqrt;
}
}
CV_Assert(_boxWidths.size() == _boxHeights.size());
_numPriors = _boxWidths.size();
if (params.has("step_h") || params.has("step_w")) {
CV_Assert(!params.has("step"));
_stepY = getParameter<float>(params, "step_h");
CV_Assert(_stepY > 0.);
_stepX = getParameter<float>(params, "step_w");
CV_Assert(_stepX > 0.);
} else if (params.has("step")) {
const float step = getParameter<float>(params, "step");
CV_Assert(step > 0);
_stepY = step;
_stepX = step;
} else {
_stepY = 0;
_stepX = 0;
}
if (params.has("offset_h") || params.has("offset_w"))
{
CV_Assert(!params.has("offset"), params.has("offset_h"), params.has("offset_w"));
getParams("offset_h", params, &_offsetsY);
getParams("offset_w", params, &_offsetsX);
CV_Assert(_offsetsX.size() == _offsetsY.size());
_numPriors *= std::max((size_t)1, 2 * (_offsetsX.size() - 1));
}
else
{
float offset = getParameter<float>(params, "offset", 0, false, 0.5);
_offsetsX.assign(1, offset);
_offsetsY.assign(1, offset);
}
}
void ExperimentalTrajectory::getDesiredValues(double time, Eigen::MatrixXd& desiredValues, Eigen::VectorXd& variance)
{
desiredValues = getDesiredValues(time);
variance = getVariance(time);
}
