Tirano System with Learning Capacity for the Detection of Moving Targets

A system based on the charge-discharge characteristics of the neural synapses of the visual path, is shortly introduced. The proposed system uses the LSR (length/speed ratio) descriptor. After image segmentation, the LSR characteristic is used to detect and classify motion in the scene. This paper mainly focuses on its simulator named TIRANO and in particular on its learning sessions and results. That's the reason why TIRANO learning capacity is demonstrated for the detection of moving targets in the following cases: (a) recognition of a family of synthetic moving elements, (b) recognition of real elements, (c) recognition of several positions of the same object, and, (d) recognition of characteristic motion situations.

• Type: Conference
• Year: 1998
• Where: NC
• Authors: Miguel Angel Fernández, Antonio Fernández-Caballero, Juan Moreno García, Gabriel Sebastián
• Keywords: LSR Characteristic, NC 1998, Neural Networks, Simulator Named Tirano, Synthetic Moving Elements