利用曲线拟合对触觉数据的压缩

Haptic Data Compression by Curve Fitting

为减轻离线触觉数据压缩带来的失真变形,提出利用曲线拟合的压缩方法.在假定触觉数据所对应原触觉信号能用分段二次参数曲线表示的前提下,通过反向采样过程来构造触觉数据拟合曲线,该曲线由分段二次曲线组成,在连接处达到切线连续,且每段二次曲线对原触觉信号具有二次多项式逼近精度.为避免保存拟合曲线的所有系数以进一步提高数据减少率,对拟合曲线系数进行了迭代求解,从而只需编码保存部分系数即可解码.实验结果表明,该方法在保持高数据减少率的前提下减轻了压缩失真变形.

In order to reduce the distortion of offline haptic data compression, a compression method using curve fitting is proposed. Under the assumption that the haptic data are sampled from the original signals that can be represented by a curve defined by piecewise quadratic polynomials, a fitting curve of the haptic data is constructed by reversing the process of generating the haptic data, which is defined by piecewise quadratic parametric curves not only with first derivatives that agree at the points where they join, but also approximating the original signals with a quadratic polynomial precision. To avoid the storage of all coefficients of the fitting curve to further improve the data reduction rate, an iterative method for solving the coefficients of the fitting curve is derived so that only partial coefficients of the fitting curve are needed to be encoded and saved. Experimental results show that the proposed method can significantly reduce the compression distortion while maintaining a high data reduction rate.