Lossy Data Compression of Vibrotactile Material-Like Textures

Tactile content will be delivered over the Internet in the near future. Vibrotactile material-like textures that resemble the surfaces of wood, leather, etc., are representative of such content. We performed lossy compression of texture data for reducing the data size. We confirmed the effectiveness of two compression strategies: quantization and truncation of data beneath a shifted perceptual threshold curve. In the quantization strategy, the amplitude spectra of vibrotactile textures could be quantized in 14 steps. This reduced the data size to approximately one quarter without any noticeable quality deterioration. The method for truncating frequency components with amplitudes smaller than a shifted perceptual threshold curve was also effective, and it was preferable to the automatic deletion of subthreshold amplitudes. We reduced the data size of vibrotactile material textures to 10-20 percent of their original size by combining the lossy data compression strategy with Huffman coding, which is a lossless data compression method. Lossy compression algorithms will enhance the online delivery of vibrotactile material-like textures by decreasing their data size without significant loss of quality.