Is there a complete guide to carrying out OBIA using OTB?

I want to learn how to carry out OBIA using OTB and I was wondering if there is a complete methodology anywhere that I can follow?

As far as I know, there is no CookBook recipe for OBIA. However, you should check the recipe about classification on vector data : Train a classifier with features

You can use QGis to compute the shape attributes on your objects (perimeter, area, …)

You can also use the new ZonalStatistics application, that compute stats of poygons over an image.

Thanks remi.cresson. But how would I use this? Can you provide a high level overview of the process?

There is not a unique path to follow.
Below is an example of OBIA workflow using only OTB:

  1. Image segmentation (For instance using the large scale segmentation mean shift),
  2. Compute features from image (spectral bands, radiometric indices, textures, etc)
  3. Compute features statistics from vector (ZonalStatistics)
  4. Train classifier (PolygonClassStatistics–>SampleSelection–>SampleExtraction–>[SampleAugmentation–>]TrainVectorClassifier, …)
  5. Apply classifier (VectorClassifier)