Unsupervised Detection and Semi-Automatic Extraction of Contours using a Statistical Model and Dynamic Programming

F. Destrempes and M. Mignotte (Canada)



In this paper, we describe a statistical model for the gradient vector field of the gray level in images based on empirical results. Moreover, we present a global Markov model for contours in images that uses this statistical mo del for the likelihood. Our model is suitable for an ICE pro cedure for the estimation of the parameters and dynamic programming (or a simulated annealing) for the segmenta tion. This yields an unsupervised method for detection of contours. Also, we present a Markov model for paths in images based on the statistical distribution of the gradient. Our model can be used to define cost functions suitable for dynamic programming as in the intelligent scissors algo rithm for semi-automatic extraction of contours.

Important Links:

Go Back