Color Image Segmentation | MRF | Potts | Gaussian likelihood | Bayesian| Simulated Annealing| python

RGB color Image Segmentation with hierarchical Markov Random Field using Potts Model, Bayesian inference with Gaussian likelihood and generalized Ising Priors with Simulated Annealing algorithm - an implementation in python

- Model-based image processing
- Scribbles used as training data, in order to compute the class-conditional means, covariance matrix and naïve priors (with MLE)
- mean IOU was used as evaluation metric for segmentation
- Simulated Annealing initialization was random
- RGB Color input image to be segmented with Bayesian MAP estimator, using Bayes Classifier (with naive class prior) and Markov Random Field (MRF) with generalized Ising (Potts) Prior (with Simulated Annealing iterative algorithm), for the data term the 2D Gaussian density was used to compute the log-likelihood of a pixel belonging to a class

#imageprocessing #imageprocessingpython #python #computervision #machinelearning #algorithm #optimization #bayesian #markov #gaussian