Enhanced genetic algorithm through scattering media by searching-strategy optimization

Abstract

Feedback-based wavefront shaping techniques are efficient approaches to compensate for multiple scattering and achieve desired focuses inside or through turbid media. In this paper, we introduced the negative linear (NL) hypothesis as the criterion to optimize the feedback-based optimization approaches, such as genetic algorithm (GA). This hypothesis reveals that the key to improving the parameters to their optima lies in acquiring maximum initial convergence speed and maintaining a negative linear decline thereafter. Based on the hypothesis, an efficient genetic algorithm based on searching-strategy optimization (GA-SSO), which applies dual adaptive control of mutation and fixed linear fitness scaling selection, was proposed to accelerate the convergence speed. The simulations and experiments confirmed that NL is a basic guide to optimize the parameters to the optima of feedback-based wavefront shaping techniques.