Creating and finding pixel values within a bounding box of an RGB image.

Hi I am looking to create a bounding box for a region of an RGB images.

I have tried things such as impoly and regionprops, both which are getting me excatly what i need, or maybe im going about it the wrong way.

I basically what the create a function where for a specific region > using a bounding box to select this area, tell me all the values of the pixels in the red, green and blue channels then the variance of them individually.

Can anyone provided code/knowledge on this approach.

Thanks,

Erin

[Walter Roberson]

When you have a bounding box (which might have been created based upon grayscale or other 2D analysis), you can imcrop() the RGB image by that bounding box. The rest is just indexing and calling var()

Answer 1

[Erin Browne]

I think I get you.

Do i greyscale the RGB images and take a bounding box of that?

If so do you know any code for doing that?

[Walter Roberson]

When you look at the color image, how do you decide where the edges of the various regions are?

[Erin Browne]

Currently I have not developed anything for edge detection as of yet,

the idea was that the change in RGB ratio between neighbouring pixels would indicate whether it is moving over say lagoon water then over corals then into the mangrove. Say the RGB ratio changed from higher in the blue to higher in the red channel this would indicate moving from lagoon water to corals, then higher in the green would be into the mangroves, do you get me?

I also have the distances in relation to how many pixels apart these regions are but I haven't yet been able to create code that can say "if RGB ratio equals (value), seven pixels over the RGB should be (value)"

I hope this makes some clarification?

[Walter Roberson]

https://www.mathworks.com/matlabcentral/answers/247003-how-to-detect-edges-in-color-image#answer_194867