如何识别不直接相互分离的轮廓？ [英] How to identify contour which are not directly separated from each other?

问题描述

请有人解释如何识别彼此不完全分开的轮廓的方形。例如，我需要识别下面图像中的方块数和它们边缘的x，y坐标。我尝试通过这个问题，但它对我不起作用。

Please can some one explain how to identify square shape of contours which are not exactly separated each other. For example I need to identify the number of squares in below image and the x,y coordinates of their edges. I try to go through this question but it didn't work for me.

所以请使用简单的代码示例来解释一下。

So please can some one explain this using simple code example.

这是我可以生成的图像，请您解释一下识别此图片中的上方方块。

This is the image that I can generated can you please explain how to identify above squares in this image.

所以请善意解释一下。

推荐答案

你必须使用事实，每个方块的红色分量等于255，并做阈值。这就是我所做的：

You have to use fact, that red component of each square equals 255, and do the threshold. Here's what I've done:

1. 进行红色分割：
   

1. Do a red color segmentation:

做扩张（去除洞）：

Do dilatation (to remove holes):

代码：

Mat src = imread("input.png"), red;
extractChannel(src, red, 2);

threshold(red, red, 254, 255, THRESH_BINARY);

Mat element = getStructuringElement(MORPH_RECT, Size( 2, 2 ), Point( 1, 1 ));
dilate(red, red, element);

这篇关于如何识别不直接相互分离的轮廓？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！