Abstract: Objective To explore the effects of different regions and amounts of fingerprint minutiae on the evaluation of likelihood ratio of fingerprint evidence， based on automatic fingerprint identification system (AFIS) score, and to provide scientific and reliable AFIS score data for the construction of likelihood ratio model for fingerprint evidence. Methods 200 fingerprints simulated to those collected at crime scene were made. According to three different minutiae regions (the center and left/right delta) and 12 different minutiae numbers ranging from 5 to 16, the fingerprints were assembled and sent to both the self-constructed same-source fingerprint database and the ten-million-person different-source fingerprint database for matching using AFIS. The match scores from the two databases were analyzed quantitatively using statistical methods. Results For different regions of fingerprints, the results including concentration trend, high score range, and maximum values of similarity scores for same-source fingerprints were significantly better than those of different-source fingerprints. Different-source fingerprints exhibited a similar data of concentration trend across different regions. Regarding the amounts of different minutia, as the amount increased, the score variation range of same-source fingerprints was larger than that of different-source fingerprints, and the dispersion of scores of same-source fingerprints gradually increased. However, the scores of different-source fingerprints exhibited different characteristics for different amounts of minutia, showing segmented situations.Conclusion This study used AFIS to conduct “black box” experiments on the region and amount of minutia influencing the evaluation method for fingerprint evidence. The quantitative analysis method is simple and easy to implement, and is also in line with the actual quantitative analysis of the factors affecting the fingerprint evidence in ultra-large-capacity databases. The results help to enrich the forensic expert’s understanding of the region and amount of minutiae, which is conducive to the construction of a scientific and reliable likelihood ratio evaluation model for fingerprint evidence.

Key words: fingerprint identification, region of minutia, amount of minutia, same-source fingerprint, different-source fingerprint

摘要： 目的 探析不同指纹细节特征的区域和数量对基于指纹自动识别系统（automatic fingerprint identification system，AFIS）得分的指纹证据似然比评价方法的影响，为指纹证据似然比评价模型的构建提供科学可靠的AFIS得分数据。方法 筛选并制作200枚模拟现场指纹，按照指纹中心和左、右三角3种不同特征区域以及5～16个的12种不同数量的细节特征进行组合，通过AFIS发送到自行构建的同源指纹数据库和千万人级的异源指纹数据库中查询比对，采用统计学方法量化分析不同细节特征区域和数量下的同源和异源比对得分。结果 对于不同细节特征区域，同源指纹相似度得分的集中趋势、高分段区间和最大值均明显优于异源指纹，异源指纹不同区域均表现出相近的数据集中趋势。对于不同细节特征数量，随着细节特征数量的增加，同源指纹得分变化区间比异源要大，同源得分的分散度逐渐增加，但异源指纹得分数据在不同数量范围内的细节特征表现出不同的特点，呈现出分段情况。结论 运用AFIS对影响指纹证据评价方法的细节特征区域和数量进行“黑箱”实验，该量化分析方法简便易行，也符合当前超大容量数据库环境下的指纹证据影响因素量化分析的实际。数据分析结果有助于丰富鉴定人对细节特征区域和数量的认识，有利于构建科学可靠的指纹证据似然比评价模型。

关键词: 指纹鉴定, 特征区域, 特征数量, 同源指纹, 异源指纹