Utility-based performance evaluation of biometric sample quality assessment algorithms

The quality score of a biometric sample is expected to predict the sample’s utility, but a universally valid definition of utility is missing. A harmonized definition of utility would be useful to facilitate the comparison of biometric sample quality assessment algorithms. This paper generalizes the utility of a biometric sample as normalized difference between the means of non-mated and mated comparison scores with respect to this sample. Using a face image data set, we show that discarding samples with low utility scores determined in this way results in a rapidly declining false non-match rate. The obtained utility scores can be used as ground-truth utility labels for training biometric sample quality assessment algorithms and for summarizing their prediction performance in a single plot and in a single Figure of merit based on the proposed utility score definition.