Application of pulse coupled neural networks in AUV’s acoustic vision system

A noval AUVpsilas acoustical vision system (AVS) model was proposed and it was composed of three kinds of high resolution imaging sonar. To better perform the tasks, we introduced the pulse coupled neural networks (PCNN) based methods for sonar image processing and analysis in the AVS. In order to reduce the noise effect, a simplified PCNN model based algorithm combined with morphological and median filter was used to wipe off the Gaussian and pulse noise simultaneously. And we also utilized the same PCNN model to get a new texture representation as the feature for seafloor sediment classification. In the formulation tests, good performances are got in the above two tasks, so it is proved PCNN an energetic processing model for the AVS in practical application.