how to find the no of colors and name of the colors in an object eg: tomato

how to find the no of colors and name of the colors in an object eg: tomato

[senthil vadivu]

thanks.But how do I differentiate the objects with similar colors by colors only

[Walter Roberson]

So for example you would like to be able to distinguish between the aqua objects and the teal objects? Or between Dusty Rose and Orange Pink?

Answer 1

[Stephen23]

Perform white/color calibration, image segmentation, and then use my FEX submission:

https://www.mathworks.com/matlabcentral/fileexchange/48155-convert-between-rgb-and-color-names

Or try some of the FEX submissions, e.g.:

http://www.mathworks.com/matlabcentral/fileexchange/28512-simplecolordetectionbyhue--

Answer 2

[Walter Roberson]

The number of colors in a picture is usually quite large. A picture of tomato could easily use a thousand different colors. Most of them have no particular name.

I suggest that you look in the File Exchange for the fuzzy color identifier. And read https://blog.xkcd.com/2010/05/03/color-survey-results/

[Stephen23]

@Walter Roberson: getting a list of color names is one thing, but matching these to RGB values is quite a different challenge. A simple RGB distance is not very good way of matching colors to names because the RGB space is very non-linear/non-uniform. That is why my FEX submission uses the common color difference methods to match RGB triples with the perceptually closest named color.

[Walter Roberson]

https://www.mathworks.com/matlabcentral/fileexchange/12326-color-name-identification--fuzzycolor

John appears to disagree in his comments

[Stephen23]

@Walter Roberson: In one comment, John talks about colorspaces Lab and HSV as being inappropriate to use as a color difference metric. I totally agree. That is why my FEX submission does not use them, and that is why I put in the effort to write code that does not use nonuniform spaces like HSV to match colors.

The Color Difference metrics that I do use are widely used: they are metrics specifically designed to allow perceptual differences between colors to be estimated. John does not mention anything about them at all, and his point about small changes in Lab directions having different effects on the color is exactly what these Color Difference metrics have been designed to solve.