Smart sensor motion detection schemes in a noisy environment

Abstract

Several motion detection schemes are considered and their responses to noisy signals investigated. The detection schemes include the Reichardt correlation detector, shunting inhibition neuron and the Horridge template model. These schemes are directionally selective and independent to the change in contrast. They essentially function by using spatial information and comparing it at successive time intervals. Using the detectors, the phenomenon of stochastic resonance (SR) is employed. SR is characterised by an improvement in response to a nonlinear system when noise is added to the input signal. Two types of SR are also considered, namely, subthreshold aperiodic and suprathreshold. Using stochastic resonance, the schemes are subjected to signals in an effort to improve the detectors responses with the addition of noise. We found that although added noise only further degrades the detectors response, an improvement can be gained by using some of the techniques from suprathreshold stochastic resonance.