Fig. 3

Flow chart of the algorithm. Initially, 20 random phase distributions are applied to the SLM. Every phase mask results in a different speckle pattern at the line shaped region of interest which is images on the camera. The algorithm evaluates the camera images. For this purpose, the ROI is divided into sub-regions as depicted in the inset. The average intensity value of the darkest sub-region of each member serves as its fitness value. The phase masks are then ranked according to their fitness values. A new population is created where the better performing masks contribute more to the phase patterns of the children. To chose the weakest sub-region as figure of merit ensures that not only the overall intensity but also the contribution of the darker parts are optimized. The algorithm can be stopped manually or it stops if the user defined maximum number of generations is created