A novel noise reduction method based on geometrical properties of continuous chaotic signals

In this paper, based on Local Projection (LP) algorithm, a new method for noise reduction in continuous chaotic signals is proposed. We use flexible curves for local approximation of the trajectory. To do so, a new adaptive algorithm is proposed to select the appropriate curve based on stretching and folding property of chaotic trajectories. Since there is a lack of appropriate index to determine the efficiency of noise reduction methods for signals, noise-free version of which we do not have, we propose a new indicator to measure the enhancement. The signals generated by Lorenz system and observed monthly series of sunspots are applied for simulation analysis and results exhibit acceptable performance of these methods.