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

An Elementary-Motion-Detector Based on Hardware Lateral Inhibition Network

Optical and Mechanical Design of a Laser-Based Non-Intrusive Vehicle Delineation Detection System

Optical and Electronic Design for a Field Prototype of a Laser-Based Vehicle Delineation Detection System

Insect-Inspired Elementary Motion Detection Embracing Resistive Memory and Spiking Neural Networks

Fully Memristive Elementary Motion Detectors for A Maneuver Prediction

Spiking Elementary Motion Detector in Neuromorphic Systems

Multiframe-integrated, In-Sensor Computing Using Persistent Photoconductivity

High‐Contrast Bidirectional Optoelectronic Synapses based on 2D Molecular Crystal Heterojunctions for Motion Detection

A Novel System for Moving Object Detection Using Bionic Compound Eyes

An insect vision-inspired neuromorphic vision systems in low-light obstacle avoidance for intelligent vehicles

All-in-one two-dimensional retinomorphic hardware device for motion detection and recognition

FPGA Design and Implementation of Insect-inspired Reichardt Motion Detector and Receptive Field

A Hardware-Based Orientation Detection System Using Dendritic Computation

A Biomimetic Visual Detection Model: Event-Driven LGMDs Implemented With Fractional Spiking Neuron Circuits

Retinomorphic Machine Vision in a Network Laser

Motion Detecting Two-Dimensional Three-Layers Analog Electronic Circuits Using Biomedical Vision Model

Optoelectronic graded neurons for bioinspired in-sensor motion perception

Mathematical study of neural feedback roles in small target motion detection

In-sensor Dynamic Computing for Intelligent Machine Vision

Computational models of object motion detectors accelerated using FPGA technology

A Universal Flexible Near-sensor Neuromorphic Tactile System with Multi-threshold strategy for Pressure Characteristic Detection