Study on compression algorithm for long-term continuous driving behavior data based on fractal interpolation

摘要

In this paper, we explore a compression algorithm for long-term continuous data based on fractal interpolation, and apply it to driving behavior data as types of respiration rate, skin conductance and blood volume plus. On basis of affine transform and compression mapping, iteration function system parameters of continuous data regarded as fractal and their data coordinates are acquired and available to be reproduced. Then, by dividing the fractal as data set into sections with the box-counting dimension value, the best compressing ratio can be speculated gradually. Before those calculations, digital filter on those electrocardiogram data such as finite impulse response (FIR) is introduced to get general outlines of those curves, and the actual unit impulse response of FIR filter is applied. Experimental data were derived from some biological signal collecting devices worn on a male test driver while he was driving a real vehicle on highways. Partial time series curves of those data are used to test the validity of the proposal algorithm. Results show that it makes the original data be compressed at least 51 times, and make sure the normed root mean square (NRMS) error is less than 0.23%. Compared with the traditional method of data compression, the algorithm based on fractal interpolation proposed in this paper is proved to be more accurate and efficient.