直方图统计中的极大值筛选

最新推荐文章于 2022-11-10 23:30:26 发布

ccss666666

最新推荐文章于 2022-11-10 23:30:26 发布

阅读量535

点赞数

分类专栏：简单算法文章标签： cv

本文链接：https://blog.csdn.net/qq_42026692/article/details/107219841

版权

简单算法专栏收录该内容

10 篇文章 0 订阅

订阅专栏

类似统计数据为0,1,1,2,3,2,1,1,0，0,2,3,3,5,3,5,3,0…等这样具有明显波峰的极值筛选。

//定义大小为Image.cols(主要避免越界，可随意定义)
std::vector<int> countPixel(Image.cols,0);
//由于可能存在相同的极值，所以对直方图数据进行平滑，权重和为1
const double filter[7] = {0.0536,0.1232,0.2032,0.2400,0.2032,0.1232,0.0536};
std::vector<double> countPixSmooth(countPiexl.size(),0);
for(int n = 3; n < countPiexl.size() - 3; ++i){
	countPixSmooth[n] = countPixel[n-3] * filter[0] 
	+ countPixel[n-2] * filter[1] 
	+ countPixel[n-1] * filter[2] 
	+ countPixel[n-0] * filter[3] 
	+ countPixel[n + 3] * filter[4] 
	+ countPixel[n + 3] * filter[5] 
	+ countPixel[n + 3] * filter[6];
}
//maxPiexl存放极大值
std::vector<int> maxPiexl;
for(int n = 3; n < countPixSmooth.size() - 3; ++n){
	if(countPixSmooth[n] >= countPixSmooth[n-1] 
	&& countPixSmooth[n] >= countPixSmooth[n+1]
	&& countPixSmooth[n] > countPixSmooth[n-2]
	&& countPixSmooth[n] > countPixSmooth[n+2]
	&& countPixSmooth[n] > 5){
		maxPiexl.push_back(n);
	}
}