Method for optimizing an optical aid by way of automatic subjective visual performance measurement

A method for optimizing an optical aid by automatic determination of a subjective visual acuity employing machine learning, the method comprising the following steps:

• a) providing at least one training data set;
• b) training an artificial neural network with the at least one training data set;
• c) generically creating a stimulus image;
• d) parameterizing the stimulus image in a first parameterization and a second parameterization;
• e) providing a recognition probability function;
• f) classifying the stimulus image by the trained neural network to provide a classification of the stimulus image;
• g) determining a recognition probability for the first parameterization performed in step d) on a basis of the classification of the stimulus image;
• h) adapting the first parameterization of the stimulus image on a basis of the determined recognition probability;
• i) classifying the adapted stimulus image by the trained neural network to provide a classification of the adapted stimulus image;
• j) determining the recognition probability for the first parameterization adapted in step h) on a basis of the classification of the adapted stimulus image;
• k) repeating steps h) to j) until a termination criterion is achieved to find a threshold for subjective visual acuity;
• l) determining the threshold of subjective visual acuity;
• m) repeatedly running through steps d) to l) with the second parameterization in each case, the second parameterization of the stimulus image in each case comprising the application of at least one property of the optical aid to the stimulus image;
• and
• n) optimizing the optical aid by selecting the second parameterization with a highest subjective visual acuity determined in step l).