Separation of Visual Scene Fragments via Oscillatory Network Method

An oscillatory network model with controllable selforganized coupling and synchronization-basedperformance was developed for image processing. Inbrightness image segmentation tasks the model demonstrates the following capabilities: a) full brightness segmentation of real gray-level and colored images; b) selective image segmentation - extraction of subset of image fragments with brightness values contained inside an arbitrary given interval. A new method of successive selection of spatially separated fragments of visual scene has been developed in the frames of the oscillatory network method based on "phase" approach. The method demonstrates satisfactory performance under some naturalrestrictions of fragment geometrical forms and theirmutual locations in visual scene.