In this article, the performance of a score-based likelihood ratio (LR) system for comparisons of fingerprints with fingermarks is studied. The system is based on an automated fingerprint identification system (AFIS) comparison algorithm and focuses on fingerprint comparisons where the fingermarks contain 6-11 minutiae. The hypotheses under consideration are evaluated at the level of the person, not the finger. The LRs are presented with bootstrap intervals indicating the sampling uncertainty involved. Several aspects of the performance are measured: leave-one-out cross-validation is applied, and rates of misleading evidence are studied in two ways. A simulation study is performed to study the coverage of the bootstrap intervals. The results indicate that the evidential strength for same source comparisons that do not meet the Dutch twelve-point standard may be substantial. The methods used can be generalized to measure the performance of score-based LR systems in other fields of forensic science.© 2017 American Academy of Forensic Sciences.

国际上,法庭科学证据的检验鉴定正朝向定量化和客观化方向转变。传统的指纹鉴定一直采用基于特征相似性比较的“认定、否定、无结论”评价模式。随着国际上对法庭科学的准确性、可靠性、客观性、透明性和可重复性等科学性要求的不断提高,指纹证据的检验评价也开始向以量化评价为核心的似然比框架模式转变。本文首先对传统的“ACE-V”指纹比对识别方法及其证据评价模式进行分析总结,然后从国际上对法庭证据技术的科学性要求出发,对指纹证据量化评价新模式的核心内容、主要方法、国际研究进展以及司法实践情况进行综述评析,重点对似然比方法体系、指纹特征的量化提取方法、特征的统计模型进行评述,最后结合指纹自动识别技术的发展,对其实践应用和发展趋势进行展望。

Forensic evidence is internationally going into quantification and objectivity with its examination and identification. Traditionally, the notion of “identification, exclusion and inconclusiveness” is the evaluation criterion about forensic evidence based on the comparison between the evidential materials from the scene and the suspect’s. With the increasing demand for the scientificity of forensic evidence and progresses of technologies, evaluation is presently paid more attentions on the eligibility for the relevant methods of examination and identification, drawing focuses on accuracy, reliability, objectivity, transparency and repeatability. Following the trend, the fingerprint evidence is also being assessed from qualitative feature similarity to the likelihood ratio quantitation. In this paper, the analytic summarization was first of all made on the traditional “ACE-V” fingerprint identification approaches and the relevant evidence evaluation framework. Successively, the new paradigms of fingerprint evidence evaluation were reviewed about the core contents, main approaches, international research development and forensic practice on the basis of introduction to the international requirements of forensic technological scientificity, having stressed into the fingerprint features that were scaled from the likelihood ratio framework, quantitative extraction methods and statistical models. Finally, the fingerprint automatic identification technology was expatiated about its development, practical application and progressive trend.

目的 研究指纹乳突纹线的一种细节特征——“间断”,考察其应用于指纹检验鉴定的价值。方法 对5000份十指捺印指纹乳突纹线细节特征进行观察统计和分析,总结“间断”特征的特性以及与其它几种常见乳突纹线特征的区别,并将该特征应用于一个实际案例中,验证效果。结果 “间断”特征具有客观存在、印痕反映较好、稳定、稀有等特性且容易辨识,可用于案件指纹检验。结论 针对特定案例,当待检指纹存在“间断”特征时,可利用该特征提高指纹检验鉴定效率。

Discontinuousness is at times observed in the friction ridge of fingerprint, thereby very potential to assist in fingerprint identification as one minutial trait. The discontinuousness is supposed to come from a big sweat pore lying on a friction ridge that has been cut into two separate parts. According to observation, there are about 24.1% showing such discontinuousness among 5000 fingerprints, demonstrating its objectivity, stability and comparative rarity. With its good moulage and recognizability, the discontinuous friction ridge can be used in fingerprint examination, helping one actual case to be solved. Therefore, discontinuousness, when found in a friction ridge of one specific case, should be not neglected but utilized.

以直接粘取法制备标准指纹,作为变形指纹研究的比较参照和依据,计算变形指纹的变形量,确定指纹变形之间存在的关联规律,即指纹在其相互垂直方向上的变形量基本相同,但以相反方式呈现。在计算机指纹自动识别系统中,若现场指纹与入库的捺印指纹越接近,其比对得分就越高;或者其细节特征数量越多,得分亦越高。但有时现场指纹因变形并不能直接与库存指纹相匹配。故在提取的现场指纹细节特征数量不能增多时,若能对现场指纹作矫正,使其提高应有的分值,则可提高比中率,增加物证的证据力。

Standard fingerprint was made by the direct picking-up and was taken as the reference and reliance for both the study and comparison of distorted fingerprint. The variance between standard fingerprint and the distorted one was calculated so as to determine the regularity among the variants. It was found that the variance is inversely equal on the mutually vertical directions of a fingerprint. For a scene fingerprint, the closer it matches to the stored one in computer automatic fingerprint identification system, and the more detailed features it discloses the higher matching score it gets. However, there are many occasions where the scene fingerprint, due to distortion, is not able to match to the one in computer automatic fingerprint identification system. Under occurrence of these circumstances plus no more available minutiae, a scene fingerprint, albeit distorted, can still be about to close to its original appearance when eliminated its variance through calibration, therefore its matching rate will be elevated and evidential power enhanced accordingly.

目的 提出一种基于Python平台的校正方法,解决曲面客体上指纹在采用光学方法提取后发生变形的问题。方法 根据标尺刻度的变化,自适应地求得最佳匹配曲面以及任意一点的放大率,通过逆变换消除曲面客体造成的畸变。结果 使用该方法校正曲率半径固定的简单曲面上指纹,结果的平均误差为5.3%,使用HGXJ-360曲面物证图像展平系统校正结果的平均误差为7%。使用该方法校正曲率半径变化的复杂曲面上的指纹,亦能取得显著的校正效果。结论 本文提出的渐变曲率曲面指纹的自适应校正方法能够自动化地校正各种曲面客体上的指纹,在简单曲面客体上其效果优于现有的成熟的校正方法,在复杂曲面客体上亦能得到优异的效果,能够为现场勘查中各类曲面物体上指纹的无损提取提供有力辅助。

<strong>Objective</strong> A Python-based self-adaptive correction was to propose for fingerprints that had been extracted from the surfaces of gradual curvature so that the difficult problem could be solved about identifying the deformed fingerprints after optical development and visualization on curving objects. <strong>Methods</strong> A calibration-engraved scaling ruler was used to measure the curvature surface where fingerprint was deposited. Based on the measurement and calculation of the deformation from the ruler&rsquo;s calibration, a correction approach was set up through Python3.4 programming plus devisal so that both the optimal matching surface and magnification rate were to figure out adaptively, making the distortion eliminated with reverse transformation. Therefore, a revision can be carried out into bringing the fingerprint to proximity to its original pattern. <strong>Results</strong> The fingerprints on simple curving radius-fixed surfaces have been able to correct with obvious effect under the average error of 5.3%, contrasting to the 7% from HGXJ-360, the Curvature-surface Physical Evidence Image Flattening System developed by Evidential Materials Authentication Center of Shanghai Public Security Bureau. For the fingerprints on complex surfaces of changing curvature radii, significant correction effects have even been attained furthermore. <strong>Conclusion</strong> The correction approach established here can automatically revise the fingerprints on various curved surfaces towards proximity to their original patterns, demonstrating effective for fingerprints on both the simple radius-fixed curvature surfaces and complex radius-changing ones, therefore capable of providing strong supports with nondestructive extraction of fingerprints on various surfaces in crime scene investigation.

我们把只出现3、4个特征的残缺指纹称为少量特征指纹。在指纹检验中,常常会遇到对只有少量特征的残缺指纹的检验。有关权威专业刊物曾报导:指纹的同一认定必须有6个以上相同的特征。由于现场条件的限制、显现固定手印方法落后、手印局部持征较少等原因,所提取的现场指纹特征出现得极少,达不到同一认定的足够数量。笔者认为,在如下特殊条件的配合下,仅凭3、4个特征相同就可以作出同一认定结论。

目的 研究硅胶仿生指纹膜印泥痕迹特征与真实指纹印泥痕迹特征的差异,为鉴定硅胶仿生指纹膜印泥痕迹提供依据。方法 首先利用液态硅胶制作仿生指纹膜,然后用仿生指纹膜和手指在相同情况下蘸取印泥,分别以轻、中、重3种力度在A4纸上垂直按压形成印泥指印,并对其拍照提取,比较仿生指纹膜印泥痕迹与真实指纹印泥痕迹的特征差异。结果 仿生指纹膜印泥痕迹多见&ldquo;空白&rdquo;&ldquo;断裂&rdquo;,边缘部分不规则、凹凸不平,小犁沟宽窄程度随力度变化明显,细节特征反映不清楚;真实指纹印泥痕迹自然、纹线较连贯。结论 仿生指纹膜印泥痕迹特征与真实指纹印泥痕迹特征之间存在差异,根据两者之间的特征差异可以帮助鉴别仿生指纹膜印泥痕迹。

<strong>Objective</strong> To discern the difference of characteristics between the imprinted marks from silicone bionic fingerprint film and real finger so as to provide a basis for identification of silicone bionic fingerprint impresses. <strong>Methods</strong> The bionic fingerprint film was made with liquid silicone, having it pressed onto inkpad along with human real finger so that such two kinds of fingerprints were made onto A4 paper where both the silicone fingerprint film and human finger were vertically impressed under three forces of small, moderate and intensive strength. The so-obtained two kinds of fingerprints were photographed and extracted, having their features and differences compared. <strong>Results</strong> The imprints from the bionic fingerprint films were mostly seen of &ldquo;blank&rdquo; and &ldquo;broken&rdquo; signs, revealing irregular and uneven edges, having various breadths of small furrows with the pressing force strength changing yet no presence of distinct nuance. Real fingerprints showed natural and coherent lines. <strong>Conclusion</strong> There are differences between the characteristics of silicone bionic fingerprint film impresses and real fingerprints, therewith capable of being used for identification of silicone bionic fingerprint impresses.

Level 3 details play essential roles in practical latent fingerprint (LFP) identification. To reliably extract reproducible and identifiable level 3 features, high-resolution images of fingerprints with adequate quality are required. Conventional methods for acquiring level 3 details often involve specific pretreatment, intricate peripheral, leading to time-consuming analysis. Herein, we simply used water to develop the sebaceous LFPs deposited on nitrocellulose (NC) membranes with only one step, and then the high-resolution (2048 pixels per inch) optical micrographs were captured to reflect the live fingertip with high fidelity. From the pictures, level 3 features, including all dimensional attributes of the ridges and pores such as number, size, location, shape, and edge contour can be extracted accurately and reproducibly. Among them, qualitative features (the structures of ridge edges) and several quantitative characteristics (the number and the relative location of sweat pores) exhibit good reproducibility. Remarkably, we proposed a new parameter termed "frequency distribution of the distance between adjacent sweat pores", short form "FDDasp", which was further proved highly identifiable in different individuals, enabling the successful distinguishment between two fragmentary fingerprints with similar level 2 structures. We believe that this methodology provides a fast and quantitative analytical paradigm for latent fingerprint identification at level 3 details.

This study addresses the permanence and persistence of friction ridges and the persistence of impressions made from these friction ridges over months and years. Permanence is the unchanging presence and appearance of friction ridge arrangements and their attributes between recurring observations of the skin. Permanence was evaluated from direct photographs of fingers collected over a period of 30-45 days (covering one or more skin regeneration cycles) as well as after 8 or more years had elapsed. Persistence embodies the operational concept of whether or not a pair of images displays sufficient similarity upon which to base an informed decision that they were made by the same finger, while acknowledging certain dissimilarities or distortions due to friction ridge physiology, image capture, matrix, substrate, and applied pressure. Persistence applies to both friction ridge skin and impressions made from these friction ridges. Permanence and persistence of skin were assessed from direct photographs of fingers taken two months apart and from finger photographs separated by an interval of at least 8 years. Permanence and persistence were also assessed from impressions taken over 4 months, as well as those separated by 8-53 years. Variability due to capture method was assessed by using four image capture methods over a four month period: direct photography of fingers, impressions captured by ink, holographic imaging, and live scan. Qualified latent fingerprint examiners assessed all changes observed over time, as well as any limitations imposed by capture method. The practice of comparison and identification of fingerprint impressions was upheld, as was the prevailing use of the word persistence to describe stability of friction ridges. All photographs and impressions of the same finger were identifiable as originating from the same source. Within all the periods of observation, level 1 detail was permanent and persistent. Persistence, but not permanence, was supported for level 2 detail. Notably, the small changes observed were only in appearance; there were no changes in the presence of new, or absence of existing, minutiae. Level 3 details of ridge edge shape and pore presence were neither permanent nor persistent. Ridge width was permanent and persistent. Incipient ridges were neither permanent nor persistent.Copyright © 2019. Published by Elsevier B.V.

Fingerprint identification errors may be due to the high similarity of fingerprints from different sources, especially when queries are conducted in a large database with the application of the Automatic Fingerprint Identification System (AFIS). In this study, a database of ten-prints of 6.964 million individuals was used; 20 sets of 60 simulated fingermarks of different qualities were used and compared with fingerprints from the database. A total of 245 queries were conducted based on both the quality of each fingermark and the number of minutiae. Four types of results were obtained from these queries on the large database, and were categorized as follows: (A) Neither Same Source nor Close Non-Match appears in the candidate list, (B) Only Same Source appears, (C) Only Close Non-Matches appear, and (D) Both Same Source and Close Non-Matches appear. When the quality of the fingermark was improved, more minutiae could be identified, and the degree of accuracy of the placement as well as orientation was higher. As a result, highly Close Non-Match fingerprints appeared; this made it harder to distinguish these fingerprints from Same Source fingerprints, especially in the large database. We concluded that more highly Close Non-Matches might appear when the database is consistently expanded, and an increasing number of Close Non-Matches might be found with a higher ranking and score than the Same Source; this would make the identification harder for examiners and might increase the possibility of identification errors.© 2021 American Academy of Forensic Sciences.

We investigated whether experts can objectively focus on feature information in fingerprints without being misled by extraneous information, such as context. We took fingerprints that have previously been examined and assessed by latent print experts to make positive identification of suspects. Then we presented these same fingerprints again, to the same experts, but gave a context that suggested that they were a no-match, and hence the suspects could not be identified. Within this new context, most of the fingerprint experts made different judgements, thus contradicting their own previous identification decisions. Cognitive aspects involved in biometric identification can explain why experts are vulnerable to make erroneous identifications.

本文从学科领域入手,对指纹自动识别技术在发展过程中受人工智能技术影响所产生的新变化进行简述。指纹识别技术作为一种计算机应用技术,其发展与计算机科学的新技术密切相关。人工智能技术,特别是基于深度学习的图像技术的发展使指纹识别算法开启了全新的模式。本文将人工智能在指纹领域的发展分成三个阶段,并对当前所处的第二阶段的发展趋势进行了分析。基于深度学习的指纹识别技术使用图像特征而不是传统细节点特征的方式,改变了法庭科学领域对指纹识别的认知。本文重点对深度学习技术在指纹识别方面的应用模式和典型的技术方法进行了论述,给出了基于深度学习的指纹识别技术方案图,对技术方案中的网络模型设计等重要步骤逐一进行了说明,提出了图像处理、降维等几个需要重点攻坚的技术环节。对现有的可为指纹识别借鉴使用的深度网络模型进行了介绍,如：卷积神经网络、自编码器网络。最后对人工智能指纹识别算法与传统算法的性能进行了对比。

An introduction was here put forward about automatic fingerprint identification relating to deep learning. Starting from the brief review on automatic fingerprint identification technology (i.e., the automated fingerprint identification system, AFIS) and developing process, the computer-based innovations were further elucidated to focus on artificial intelligence (AI) technology and its application in recognition of images. Fingerprint identification is definitely a kind of image discrimination, hence having it brought into a new way of AI technology tackling. With the recognition algorithm about fingerprint pattern, AI is playing its role in fingerprint identification through improving image technology based on deep learning. AI&rsquo;s fingerprint identification can be divided into three stages, leaving the second stage being here specifically described about the relevant developing trend. Forensic fingerprint recognition has since been changed with deep learning adopting image features other than traditional minutiae distinctions. The emphasis of this essay was put on the deep learning technology about its application mode and typical technology method in fingerprint identification, with the demonstrative technical scheme being illustrated. For the involving network model design in the technical scheme mentioned above, those important steps were explained one by one. Several key technical problems were also proposed about image processing, dimension reduction and the others related. The present available deep learning network models, eligible for fingerprint identification, were introduced with the examples of convolutional neural network and the auto-encoder one, too. Finally, the comparison was carried out on performance of AI&rsquo;s fingerprint identification algorithm against the traditional one.

本文旨在总结指纹技术的新发展,并从以下10个方面分析指纹技术未来发展的可能方向。1. 更灵敏的显现试剂。以荧光试剂和纳米粒子为代表的两种技术最为重要和突出。纳米粒子较常规粉末吸附性好,而荧光试剂具有高灵敏度且能克服背景干扰的优点,二者结合表现更加显著,有关研究表明其前景广阔。2. 疑难客体上的指纹显现技术。指皮肤、胶带粘面、塑料纸币及背景发荧光的材质等表面上的指纹显现。3. 时间分辨和相分辨技术。它们都涉及到一系列复杂仪器的使用,另外与传统荧光使用不同,时间分辨技术是根据指纹试剂及背景的荧光寿命或者相位的不同,通过复杂仪器分辨出该微小差异并加以放大,从而将指纹显现出来。此方法能显现传统荧光法处理不了的指纹。4. 光谱成像技术。包括红外、紫外及可见光、拉曼成像等。光谱成像技术特别是红外光谱成像技术在显现指纹的同时,能够分辨出指纹物质的成分,比如手上粘附的一些外源性物质像毒品、爆炸残留物等。因而,光谱成像技术可能是迄今为止最为有效的能解决一些疑难指纹显现的技术,但光谱成像一般需要大型昂贵的仪器设备。5. 生化核危害性物质污染的检材上的手印显现。甲醛处理生化类污染的指纹检材有报道,但核污染材料本文未涉及。6. 免疫学和适配体技术。使用抗原抗体的免疫学反应来显现指纹也是人们探索的重要方向之一。免疫学和适配体技术都具有高灵敏度和选择性强的优点,但反应条件较为苛刻。7. 指纹来源的情报信息获取。从指纹中探测毒品、爆炸残留物等信息属于指纹信息学的范畴。此外,指纹自动识别系统的指纹信息也是侦查破案所需的重要情报信息。8. 指纹鉴定及三级特征的应用。现有指纹鉴定是以二级特征数量为标准的,但在实际案件中,常有二级特征不足的情形,汗孔及指纹纹线微小形状等三级特征可作为重要辅助特征而帮助鉴定。9. 指纹遗留时间的判断。该技术和方法具有重大意义,但也一直存有难点。有报道通过测定指纹遗留物质中棕榈酸的扩散速度,初步确定出其与指纹遗留时间的相关性。但影响判断指纹遗留时间的因素太多,建立通用的指纹遗留时间判断模型仍需艰苦的工作。10. 计算机指纹自动识别技术（AFIS）。AFIS在中国发展很不均衡。国家层面没有统一的AFIS,而是由各省分别建立,这导致了指纹工作总体效率较低。为此,公安部建立了指纹协查平台,出台了系统认证等办法,部分解决了既有难题。指纹大库建设现正在准备和论证中。

This article tries to summarize the recent advances of fingerprint technology and demonstrates ten possible developing directions in the future: 1. more sensitive reagents; 2. fingerprit detection on surfaces difficult to handle; 3. time-resolve (TR) and phase-resolve (PR) technology; 4. chemical imaging technology; 5. fingerprit detection on the exhibits polluted by bio-, chem- and/or nuclear-hazardousness materials; 6. immune and aptamer technology; 7. forensic intelligence from fingerprit detection; 8. the use of 3rd level characteristics in fingerprint identification; 9. age estimation for fingerprits; 10. more powerful Automatic Fingerprint Identification System (AFIS).