A Model Study Of The Response Under The Noise Background Of Motion-Sensitive Neurons In The Optic Tectum Intermediate And Deep Layer Of Pigeon

Birds have a well-developed visual system,which can identify and track targets on the ground when flying at high altitudes.The optic tectum is an important nucleus in the visual pathway from the tectum,which is the key to realizing this ability.Therefore,analyzing the mechanism of optic tectum neurons perceiving small targets under the noisy background and constructing a corresponding coding model have important application significance for deepening people’s understanding of the ability of biological perception of small targets,and for forming small target detection algorithms in noisy environments.In this paper,taking the optic tectum,an important nucleus of the tectal pathway,as the research object,electrophysiological experiments were carried out on the motion-sensitive neurons in the intermediate and deep layers of the optic tectum with a depth of 1000-1300μm,and the corresponding neural signals were collected and analyzed.The mechanism of continuously moving weak and small targets under the noisy background is perceived,and a target location decoding algorithm based on the response of motion-sensitive neuron clusters is presented.The main work and research results of this paper are as follows:(1)The response properties of motion-sensitive neurons to small moving targets under noise background are analyzed.Designing different intensities of salt-andpepper noise and natural scenes with local micro-movements as the background of moving target stimulation paradigms,and collecting neurophysiological signals of such neurons,it was found that motion-sensitive neurons in the process of perceiving small moving targets has a certain degree of robustness to interference of the scene.(2)The encoding model of motion-sensitive neuron sensing small moving objects in the background of noise is constructed.Based on the sequential probability activation hypothesis and the directional energy accumulation hypothesis,combined with the electrophysiological characteristics of neurons to small targets moving continuously under the background of salt-and-pepper noise and local fretting in natural scenes,the hypothesis of the maximum reward value criterion and the proximity criterion assumptions of the dendritic field are proposed;the model simulation results of the two hypotheses and the electrophysiological results were compared and analyzed,and the results showed that the maximum reward value criterion assumption is more in line with the guess of the neuron’s ability to resist noise interference;finally,combined with the assumption of the maximum reward value criterion,an encoding model of small moving objects under the background of noise perception by motion-sensitive neurons is constructed.(3)A target location decoding algorithm based on the response of motionsensitive neuron clusters is presented.The location information of the target is decoded by the cluster response of the motion-sensitive neurons to the moving small target in different sport modes under the background of salt and pepper noise and the background of natural scene.The decoding results verify the credibility of the maximum reward value criterion assumption and the validity of the decoding model.In this paper,by analyzing the anti-noise interference characteristics of motionsensitive neurons,the hypothesis of the maximum reward value criterion is proposed,and the coding model of motion-sensitive neurons to perceive small moving objects in the background of noise is constructed,and a target position decoding algorithm based on the motion-sensitive neuron cluster coding model is given.This research not only deepens people’s understanding of the ability of biological perception of small targets,but also has important application significance for the formation of small target detection algorithms in noisy environments,and is expected to form a new brain-like computing system different from deep learning.