摘要 |
The device for the segmentation of data includes an initial segmentation categorising sub-assembly (100) which provides an initial set of labels. A second sub-assembly (200) then determines conditional probabilities, based on a propagation operator (L) whose parameters are defined by given relationships. This is achieved in two parallel channels. The probabilities are used in a labelling sub-assembly (300) which provides a new set of labels. Finally, a sequencing sub-assembly (400) controls, at the input of the labelling sub-assembly (300), the replacement of the initial set of labels with the new set. This sub-assembly also determines whether the iterative procedure for the determination of new labels should be continued or discontinued. |