how to classify imageData Store images depending on each image's object shape

I have ImageDataStore that has 1000 images and looking to classify each images depending on the shape of the object inside each image instead of classifying depending on RCS

Answer 1

[Image Analyst]

See my attached shape recognition demo. If you need more help, attach your images and say what shape(s) you want to find in them, AFTER you read this:

TUTORIAL: How to ask a question (on Answers) and get a fast answer

[Mohamed Elbeialy]

Thanks for your help, however, your code does not match my question. In particular, I do have an imageData store that includes images like the attached ones. All what I need, to classify those images inside the imageData Store depending on the shape of each object in each image( finding the max value), rather than using RCS info of each image.

[Mohamed Elbeialy]

Did you get a chance to check my question ?

[Image Analyst]

I'd probably threshold to get the flower and get a mask of the largest blob. Determine the thresholds by analyzing the colors around the edge of the image and in the central part and finding the color half way between. OOr ir all your flowers are red, just take the red channel and call imbinarize, followed by imfill() and bwareafilt() to get the largest blob. Then I'd get the boundaries with bwboundaries() and find the disances of each boundary point ot the centroid. Do this for each flower to make up a reference library of distance signals. Then I'd use normalized cross correlation with the distance signal from an unknown flower to see which reference flower it matches up best with.

[Mohamed Elbeialy]

Steps might be confused for me. Could you help my with the code. Remember that this is an imageData store and not just the pics showed above. So, I want to make sure that the process of detecting the shape by finding the max value in each image will be done for all images inside the Datastore, then classify and train them through deep learning network

[Image Analyst]

Not sure I'll have time. What's your "use case"? Why do you need to do this? And where/how are you getting the photos.

[Mohamed Elbeialy]

It's a project and all what I need is to find the max value of all images inside the Datastore to classify by object's shape inside each image instead of RCS. I do not need the whole code for classifing, only the part of finding the max value