I am trying to classify 9 types of crops (Y) from 5 satellite images (X), acquired at dates t = {t1,...,t5}, corresponding to 5 crop phrenology states (S) i.e. preparation, seeding, growing, harvesting and post-harvesting stages. I am using similar training parcels throughout therefore the prior probabilities (initial probabilities) are equal. I have computed the likelihood model (Emission Probabilities) P(X|Y) from each image using Gaussian mixture model and I have state Transition probabilities for each crop. My question is how do I integrate this into HMM using this library to determine the final class of each pixel {using HMM posterior probability model equation below} in a final image of the area? So far I have looked at the example used here here but it is still not clear for me.
HMMM model equation:
Kindly advice. regards.
