Image segmentation with the spatiotemporal neuron and adaptive threshold learning

It is shown that a network of SpatioTEmporal Neurons (STEN), in combination with Adaptive Threshold Learning (ATL), can be used to segment images. The network performs the segmentation by temporally binding regions in the image with similar characteristics. Further, with proper selection of parameters, it is possible to extract related features such as edges and corners of regions. The authors have applied this method to segmenting a portion of an MRI image.