Automated fingerprint identification system (AFIS) benchmarking using the National Institute of Standards and Technology (NIST) Special Database 4

The recent growth in the number of products that use fingerprint matching technology has spawned a need for appropriate evaluation methods. We propose a set of benchmarks based on NIST Special Database 4 to support the evaluation of fingerprint matching technology. The approach used to develop the benchmark parameters and performance measurement methods are discussed. Benchmark parameters were established with members of the law enforcement community. Characteristics that typically affect fingerprint matching algorithm performance are observed and measured. Variation of matcher performance due to these characteristics, and others, indicates a need for standard data (NIST Special Database 4) when comparing algorithm performance. Benchmark parameter measurements using the United Kingdom Home Office HO-39 fingerprint matching algorithm and a Sun SPARC 10 are provided. Analysis of benchmark results provides insight into matcher performance. An example demonstrates how the benchmarks may be used