Biological weight selection of multi-scale retinex via artificial bee colony algorithm

摘要

The goal of this study is to present a biological approach to weight selection in Multi-scale Retinex (MSR) using the Artificial Bee Colony (ABC) algorithm. The standard MSR assigns the same weights to the Gaussian filters with different scales and cannot ensure the optimal enhancement results in various environments. To tackle this problem, we employ ABC for weight selection to optimize the evaluation results in MSR. The optimization of weight selection compensates the defects of the scale parameters in Gaussian filters. Some examples are given to demonstrate the feasibility and potential of the approach. We report our experimental results to demonstrate the excellent performance in enhancement compared with standard MSR. The advantages of our method are systematically analyzed in detail.