Automated Color Extractor - TinEye Labs Color Extraction

1 次查看（过去 30 天）

Kent Yeap 2017-3-19

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/330715-automated-color-extractor-tineye-labs-color-extraction

关闭： John D'Errico 2017-3-19

Hi, I would like to develop an automated color extractor which is exactly like this http://labs.tineye.com/color/. Results of the color extraction are as shown below. The automated color extractor is able to automatically identify number of dominant colors (i.e., only one color, 2 or more colors) in the image, then show the color map of the image, followed by its hex color, color area and closest color name.

Based on some online research, I understand that K mean clustering is required to segment image into different clusters, however, the K mean clustering method that I found requires manual user input of number of cluster (i.e., user need to tell the number of cluster required), hence, some modification are required so that it's automated.

Here's are some of my questions:

1. How can I automate the K mean clustering so that it's able to identify the number of clusters automatically?

2. What method is required to produce such color map where the image is smoothen and palette color is shown?

3. How to identify the closest color (e.g.: mandarin orange, bright grey, air force blue) and its base color (e.g.: red, grey, blue)?

Any links to similar/existing projects or guidance is much appreciated! Thank you in advance.