Fig. 2.

A comparison of the MSE loss functions for the GA and CMA-ES optimization methods. Additionally, five images at different iteration numbers (10, 20, 30, 50 and 73) for the CMA-ES method as well as the target image (T) are shown. Note: Images represent one sample image out of a population size of four acquired with various parameter combinations. As a result of the stochastic feature of the GA, the intensity distribution of image 3 is not between image 2 and image 4. In contrast, MSE loss represents the average loss of all four images.