Periodic pattern enhancement: A stochastic resonance approach

This paper investigates two basic problems about the periodic pattern in pattern recognition and artificial intelligence: weak periodic pattern and the noise. We make use of stochastic resonance (SR) theory to enhance periodic pattern by transferring the noise energy to the pattern energy, which not only removes the noise but also improves the periodic pattern intensity. We firstly analyze the basic principle of the energy transfer by SR theoretically; and then, we design the optimized SR nonlinear system to enhance the periodic pattern; finally, we apply the proposed method to two popular periodic patterns, i.e., the texture pattern of the optical remote sensing image and the spatial pattern of the fingerprint image. The experiments have a good performance and the proposed method can be extended to wider applications of periodic pattern enhancement.