I have extracted all the size blobs form a dataset.

using the K-means i am trying to classify theses blobs into 2 classes(vehicles,pedestrians). how i could using the area blobs to do this.

[Image Analyst]

What are the features you're measuring and going to pass to kmeans()?

[sassi nizar]

i have divided the image into blocks of 20*20, and i wanna calculate the size of the object in each block. After that, i pass the size and cell(blocks) to K means.

[Image Analyst]

That will produce nonsense.

Answer 1

[KJVKU999]

1. Decide first your features
2. Prepare features from your extracted blob/blob size data, and
3. You need to make table/matrix of the features including the class name/label.

• Now pass your feature matrix/table to one of the classification methods. The procedure will generate one structure file from which you can predict un-labled/unknown data.

[KJVKU999]

If Area of the blob is your feature:

• Calculate area of the blobs and prepare matrix/table. Follow above steps.