Synchronised oscillator network
Network of synchronized oscillators [10] is a recently developed tool is based on "temporary correlation" theory, which attempts to explain scene recognition as it would be performed by a human brain. This theory assumes that different groups of neural cells encode different properties of homogeneous image regions (e.g. shape, color, texture). Monitoring of temporal activity of cell groups allows detection of such image regions and consequently, leads to scene segmentation.
Oscillator network was successfully used for segmentation of textured biomedical images [1,2,4]. The advantage of this network is its adaptation to local image changes (related both to the image intensity and texture), which in turn ensures correct segmentation of noisy and blurred image fragments. Another advantage is that synchronized oscillators do not require any training process, unlike the artificial neural networks.
Oscillator network is also able to detect texture boundaries, which provides much faster image segmentation (number of excited oscillators is reduced to those located on texture boundary only [3]. This network performs also morphological filtering [5] and can be applied for pattern recognition [8].
Finally, such a network can be manufactured as a CMOS VLSI chip, for very fast image segmentation [9].
Fig. 1. MRI human feet cross-section (a), results of heel bone detection (b)
Fig. 2. Echocardiogram with benign heart tumor (a), tumor segmentation results
Fig. 3. Detection of texture boundaries for images from Fig. 1a (a) and Fig. 2a (b)
Fig. 4. Segmentation result for 3D human liver images
