Portal de Periódicos da CAPES

Dual‐layered oscillatory model for object detection and tracking

2017; Institution of Engineering and Technology; Volume: 53; Issue: 16 Linguagem: Inglês

10.1049/el.2017.1788

1350-911X

You Mi Hwang, Mungwi Jeong, Won Jin Cho, Sang–Rok Oh,

Neural Networks and Applications

A method to detect and track objects using an oscillatory neural model is presented that mimics the integrative component from the primary visual cortex to the vision-related parietal and temporal cortex. The locally excitatory globally inhibitory oscillator is incorporated into the proposed model to implement synchronisation and desynchronisation of neural oscillation, and the dual-layer architecture (composed of the form layer corresponding to the ventral pathway and the motion layer to the dorsal pathway) is also introduced to implement the integrated pathways of the human visual process. Objection detection corresponds to a function in the ventral pathway, and tracking of the detected object corresponds to a function in the dorsal pathway. Some experiments where skin regions were detected and tracked are carried out, and showed that the proposed model of the integrative pathways in the human visual process works successfully.