Vibrotactile Signal Compression Based on Sparse Linear Prediction and Human Tactile Sensitivity Function

Abstract

In this work, we present a novel vibrotactile coding scheme that encompasses a sparse linear predictor and a perceptual compressor. The predictor introduces a sparsity constraint both on the prediction coefficients and the residual. The prediction residual is then filtered and quantized using a human tactile sensitivity function generated from the vibrotactile detection threshold-frequency characteristics. Furthermore, a novel objective quality assessment method (ST-SIM) for vibrotactile signals that embraces perceptual spectral and temporal similarity measures is developed. ST-SIM is then used to evaluate and validate the overall signal quality and proposed compression scheme performance using different vibrotactile signal contents.