Neuromorphic visual target classification method and system

Abstract

The invention relates to a neuromorphic visual target classification method and system. The method comprises the following steps: acquiring a space-time pulse event stream asynchronously output by anevent camera; determining a time surface of each time-space pulse event in the time-space pulse event stream according to the time-space pulse event stream; determining a time surface prototype according to the time surfaces of the first K time-space pulse events in the time-space pulse event flow; updating the time surface prototype by using the time table surface of the residual space-time pulseevent in the space-time pulse event flow, and constructing a hierarchical model of the time surface; and classifying the spatial and temporal features by adopting a single-layer spiking neural network of a group coding Tempotron neuron. The invention provides a neuromorphic vision target classification method and system, and effectively solves the problem of target identification and classification of neuromorphic vision.

Description

technical field [0001] The invention relates to the field of object recognition and classification of neuromorphic vision, in particular to a method and system for classifying objects of neuromorphic vision. Background technique [0002] Brain-inspired intelligence is an important development direction of artificial intelligence in the future. Its goal is to establish a brain-inspired artificial intelligence system with the same efficiency as biological computing by drawing on the neural structure and information processing mechanism of the brain, so as to reach or surpass the level of human intelligence. With the rapid development of artificial intelligence, the fields of brain-inspired research and neuromorphic computing have attracted extensive attention from researchers at home and abroad. By simulating the working mechanism and principle of the biological retina, neuroscientists have developed a silicon retina device, a neuromorphic vision sensor. Unlike traditional vi...

AI Technical Summary

Problems solved by technology

In the existing technology, the Gabor filter can be used to realize the feature extraction part. However, the robustness of this method to noise is relatively poor, and additional data preprocessing is required for noise reduction to improve the accuracy of recognition and classification.

In addition, Gabor filters are used to define different strip feature templates, which are usually sensitive to the noise of data samples and have weak generalization ability.

Images