颜色直方图归一化 [英] Color Histogram Normalization
问题描述
昨天我在这个问题上找到了这个问题:
直方图规范化。
I asked this question yesterday on this thread found here: Histogram Normalization.
但是,我尝试使用链接中的一些想法对图像执行标准化,并得到以下结果。在这里,我将所有像素标准化为50%,黑白像素除外。
我使用的公式是:(像素 - 分钟)/(最大 - 分钟)* 127
However, I tried using some ideas i got in link to perform Normalization on a image and i got the following result. Here I normalize all the pixel to 50% except the black and white pixels.
The formula i used is:(pixel - min) / (max - min) * 127
pixel = (float)src.at<uchar>(j,i);
if (pixel == 255)
{
img.at<uchar>(j,i) = pixel;
}
if (pixel == 0)
{
img.at<uchar>(j,i) = pixel;
}
/*if (min == 0 || max == 0 || (max - min == 0))
{
img.at<uchar>(j,i) = pixel;
}
else
{*/
normal__ = ((pixel - min)/(max - min)) * ( 127);
img.at<uchar>(j,i) = normal__;
//}
}
结果:
RED PIXEL: MIN = 0 MAX = 253
GREEN PIXEL: MIN = 0 MAX = 254
BLUE PIXEL: MIN = 0 MAX = 255
标准化前的图像
Image before Normalization
使用上述公式进行标准化后:
After Normalization using above formula:
现在,我确认我的步骤是否正确是正确的。谢谢.. :)
Now, I am confirming if my steps are correct. Thank you..:)
推荐答案
因为陈述
normal__ = ((pixel - min)/(max - min)) * ( 127);
img.at<uchar>(j,i) = normal__;
不在if-else语句中,它们也适用于黑白像素。那没关系吗?
are not in an if-else statement, they get applied to the black and white pixels too. Is that fine?
这篇关于颜色直方图归一化的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持IT屋!