使用自组织图进行图像分割 [英] Image segmentation using Self-Organizing Maps

问题描述

您好，我想使用SOM分割图片.我不知道如何生成输入向量.我知道我需要从图像像素中提取像素，我的问题是如何手动指定像素的类?
抱歉，我的英语不好.

Hi, I want to segment an Image using SOM. I have no idea of how the input vectors are generated. I understand I need to extract that from image pixels, my question is how do I designate the pixel''s class?Manually?
By the way,sorry for my poor English.

推荐答案

您正在使用哪个工具?通过所使用的库输入更改的方法.

通常，最简单的是，存在一个Nx3矩阵，用于定义特征向量，例如:

R1 G1 B1
R2 G2 B2
...
...
...
Rn Gn Bn


结果将是Nx1向量，即所有像素的类别标签.

此方法不限于SOM.它适用于许多无监督的分类器，例如GMM等.

最好，

Hi,

Which tool are you using? The way to input changes by the library you are using.

Generally and most simply, there is an Nx3 matrix, defining the feature vectors, such as:

R1 G1 B1
R2 G2 B2
...
...
...
Rn Gn Bn


The result will be an Nx1 vector, a class label for all the pixels.

This method is not limited to SOMs. It''s valid for many unsupervised classifiers such as GMMs etc.

Bests,

这篇关于使用自组织图进行图像分割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！