Method and device for automatic image segmentation by textural analysis

摘要

A device for automatic image segmentation according to the invention comprises a data processor associated with a network of automata organized in three levels; the inputs of each of the automata are weighted by adjustable coefficients; the number of automata of the first level is equal to the number of characteristic values computed from observation windows taken firstly in a set of examples of image zones having different textures. Learning by the network permits adjustment of the weighting coefficients of the automata of the network until all the examples of the same texture lead to the same configuration.