Physiological Tremor Filtering without Phase Distortion for Robotic Microsurgery - Matlab implementation codes
The zip folder, RSSA-ELM algorithms.7z consists of three Run codes - Run_RSSA_ELM.m, Run_MW_RSSA_ELM.m and Run_Enhanced_MW_RSSA_ELM.m. These codes execute RSSA-ELM, MW RSSA-ELM and Enhanced MW RSSA-ELM algorithms respectively. The algorithms validate their performance with a synthetic data, synth_data2.mat. This synthetic data simulates the real hand recordings used in the paper, and it includes a mixture of synthetic voluntary motion, tremulous motion and low frequency noise. The data is 12s long, using the first 4s for the offline SSA estimation, RSSA-ELM algorithm estimates and predicts the remaining 8s of the data sample-by-sample. Similarly, MW RSSA-ELM and Enhanced MW RSSA-ELM algorithms each has a moving window of the fixed 4s length. These algorithms perform estimation and prediction by shifting this moving window by one sample at each iteration, as described in the paper. Only 4s of window length is chosen, as the representative data used in this demo is short, the window length can be adjusted as required to obtain the desired performance.
All the algorithms perform 10 sample ahead prediction, providing the final estimation with 9-sample delay compared to the latest input sample.