Mathematical study of neural feedback roles in small target motion detection

• DOI: 10.3389/fnbot.2022.984430
• Publication Date: 2022-09-20T08:23:06Z
• Authors: Jun Ling; Hongxin Wang; Mingshuo Xu; Hao Chen; Haiyang Li; Jigen Peng
• Keywords: Motion Detection

Building an efficient and reliable small target motion detection visual system is challenging for artificial intelligence robotics because a only occupies few pixels hardly displays features in images. Biological systems that have evolved over millions of years could be ideal templates designing systems. Insects benefit from class specialized neurons, called detectors (STMDs), which endow them with excellent ability to detect moving targets against cluttered dynamic environment. Some bio-inspired models featured feed-forward information processing architectures been proposed imitate the functions STMD neurons. However, feedback, crucial mechanism regulation, has not investigated deeply STMD-based neural circuits its roles remain unclear. In this paper, we propose time-delay feedback model complex backgrounds. The main contributions study are as follows. First, pathway designed by transmitting output-layer neurons lower-layer interneurons role analyzed view mathematical analysis. Second, estimate constant, existence uniqueness solutions nonlinear dynamical formed loop via Schauder's fixed point theorem contraction mapping theorem. Finally, iterative algorithm solve problem performance tested experiments. Experimental results demonstrate able weaken background false positives while maintaining minor effect on targets. It outperforms existing regarding accuracy fast-moving clutter. approach inspire relevant modeling robust perception