高斯背景建模学习之icvUpdateFullWindow函数篇

最新推荐文章于 2018-01-23 00:57:15 发布

读书与远方

最新推荐文章于 2018-01-23 00:57:15 发布

阅读量876

点赞数

分类专栏：图像处理机器视觉 Opencv

本文链接：https://blog.csdn.net/yingchunhua365/article/details/16339103

版权

图像处理同时被 3 个专栏收录

62 篇文章 2 订阅

订阅专栏

机器视觉

41 篇文章 0 订阅

订阅专栏

Opencv

21 篇文章 0 订阅

订阅专栏

static void icvUpdateFullWindow( double* src_pixel, int nChannels, int* match,
CvGaussBGPoint* g_point,
const CvGaussBGStatModelParams *bg_model_params )
{
const double learning_rate_weight = (1.0/(double)bg_model_params->win_size);
for(int k = 0; k < bg_model_params->n_gauss; k++)
{
//这个就是更新权重，w=(1-a)w+a*m (a:更新率，m是匹配，匹配就是1，不匹配就是0)
//可以改写：
// g_point->g_values[k].weight = g_point->(1-learning_rate_weight)*g_values[k].weight
// +learning_rate_weight*(double)match[k]
g_point->g_values[k].weight = g_point->g_values[k].weight +
(learning_rate_weight*((double)match[k] -
g_point->g_values[k].weight));
//更新方差，均值等参数
if(match[k])
{
//learning_rate_gaussian实际就是p（在更新方差和均值要用的参数）
//实际上可以这样设置：
//double learning_rate_gaussian = ((double)match[k]*learning_rate_weight)/(g_point->g_values[k].weight
double learning_rate_gaussian = (double)match[k]/(g_point->g_values[k].weight*
(double)bg_model_params->win_size);
for(int m = 0; m < nChannels; m++)
{
//均值和方差的更新公式
//u = (1-p)*u+p*x = u+p*(x-u)
//o = (1-p)*o*o+p*(x-u)*(x-u)
const double tmpDiff = src_pixel[m] - g_point->g_values[k].mean[m];
g_point->g_values[k].mean[m] = g_point->g_values[k].mean[m] +
(learning_rate_gaussian * tmpDiff);
g_point->g_values[k].variance[m] = g_point->g_values[k].variance[m]+
(learning_rate_gaussian*((tmpDiff*tmpDiff) - g_point->g_values[k].variance[m]));
}
}
}
}