Improving Accuracy and Effectiveness of Forensic Ballistics Analysis & Technology

8.3 Improving Accuracy and Effectiveness of Forensic Ballistics Analysis & Technology

University of Huddersfield and Forensic Pathways Limited

1. 
   Abstract
   

In recent times automation has industrialised this process by the use of machine assisted methods to automate part of the process. Advances have delivered the ability to acquire and store digital images of ballistic exhibits, store them in a database, undertake correlations to find potential matches and report these to an expert. At the end of the process the firearms expert will be presented with a list of potential matches. But inevitably the expert is faced with a decision about declaring the most probable match based upon a range of potential candidates. Interestingly in order for the expert to declare an evidential match there is the need to resort back to conventional microscopy. (Bachrach, 2006).

Whilst the current technology has resulted in an apparent speeding up the forensic process, new problems have been created for example;

• 
  Data acquisition process has not been standardised to a robust control model
  
• 
  The correlation algorithms are operating across noisy data
  
• 
  The different technologies lack interoperability so cross comparison is not possible
  
• 
  There are no global standards (including statistical) for the declaration of a match
  

1. 
   Introduction
   

1. 
   Ancillary crime data routinely collected by law enforcement organizations is included in the forensic process;
   
2. 
   Analysis of ballistics data and crime information is systemized and semi-automated;
   
3. 
   Data sharing is adopted on a wide-scale;
   
4. 
   Data capture techniques are improved;
   
5. 
   Technical interoperability of data and systems is achieved.
   

This paper also discusses future state of the art in ballistics analysis technologies and the likely impact on the detection and prevention of gun crime.

Forensic ballistics examinations are based upon the principle that physical impressions from the firearm are left on fired bullets and cartridge cases when the weapon is fired. This is called the “Principle of Exchange” (Locard) and is based upon his theory that whenever two objects come into contact there is an inevitable exchange of material between the objects. Additionally, we postulate that whenever two objects come into contact with each other there is an equal and opposite exchange of force. The resulting impressions can be indicative of the method and type of contact. If these impressions can be recorded and preserved they can be compared with other impressions on other objects that occurred at different times but with similar results. Detecting evidence of the exchange forms the basis of the modern forensic ballistic process.

Forensic firearm and tool mark examinations involve analyzing firearms, ammunition and tool mark impressions to establish whether a certain firearm discharged a particular bullet or cartridge case or whether a tool was used in the commission of a crime. The basis of the methodology in both cases is the same. A recovered object whether a bullet, cartridge case or a tool mark impression, can be compared with any other sample of a similar type to ascertain if there is a correlation between the physical marks left on or by both objects. Under microscopic conditions, these comparisons can be highly discriminating. However, this process is made difficult because:

• 
  No two objects are exactly the same (identical) but may be very similar. (De Kinder, 2002)
  
• 
  Firearms wear over time resulting in different marks (De Kinder, 2002)
  
• 
  The marks left on a cartridge case or bullet are slightly different every time a gun is fired (De Kinder, 2002). The physical signatures of Firearms therefore evolve over time and with use
  
• 
  Ammunition type directly affects the quality of marks (De Kinder, 2002)
  
• 
  Some weapon types leave marks that are particularly difficult for humans to distinguish between. An example of this are Glock weapons. (Bachrach, 2006).
  
• 
  The examiner has to distinguish between class characteristics typical of the make and model of weapon and individual characteristics unique to a particular weapon. (Saks and Koehler, 2005).
  

Once two objects have been matched together, inferences can be made, hypotheses generated and conclusions drawn about relationships and links between crimes, suspects, weapons and the many different types of associated evidence. (Leon, 2006).These links can reveal important information that can help with the detection of a crime and also provide additional intelligence to law enforcement agencies. The nature of this process is founded upon predicate logic and this forms the basis of another Paper currently awaiting publication.

2. 
   Current State of the Art
   

The resulting images from all these systems can reveal markings that are specific to a particular make and model of weapon as well as marks that are specific to a particular firearm. Databases of these markings have been developed and are in use.

Each type of ballistics system works in a different way and utilises different methods. These methods are the intellectual property of the organisations and inventors concerned. However, all use light sources directed at the object but each uses a different angle or angles, different lighting mechanism, different capture technique and different levels of light intensity. Different algorithms are applied to the images to extract data, store it and undertake correlations. The methods used to perform correlations also use different algorithms. The important point is that the underlying principles of data capture and correlation are similar and therefore offer the theoretical possibility of interoperability. It is useful to consider the process as having three steps:-

1. 
   Data capture (Collect and store)
   
2. 
   Data indexing or feature extraction (Analysis of features and storage)
   
3. 
   Correlation techniques (analysis of features in relation to features extracted from other objects)
   
4. 
   Declaration of potential matches (Grouping of Potential Matches)
   

The nature and characteristics of the marks observed on objects are such that each technology currently can only produce a candidate match list that contains objects that the system suggests are similar to the object under examination. These lists vary in composition and length depending on the system being used and its’ configuration. In order for a match to be declared and presented as evidence in court, a ballistics examiner has to visually examine the objects using conventional microscopy. The candidate list is produced as a result of the database providing those objects which may bear similar marks to the questioned object under examination. The examiner is thus presented with a pre-determined population of objects for comparison with the questioned object. This research and our contact with experts to date indicates that the object selected as the most likely match by the machine is sometimes not the selected evidential match by the examiner when visual and confirmatory examination takes place. This raises questions about the need for optimisation.

Current state of the art approaches can result in inferences that weapons and objects are linked. This can then lead to inferences that crimes, people, firearms and associated evidence is linked. Table 1 shows how links between cases can be inferred. The Matrix is a representation of a database where information concerning gun crime cases is stored. In this example there are five cases. Our experience to date suggests that the current state of the art almost always results in links between small numbers of cases and typically one to one case matches. Our experience in other areas of informatics leads us to believe that this is due to the lack of access to data and systems together with the lack of use of ancillary crime data.

Returning to the current state of the art, the following “Ballistics Matrix” and “Key” provides details of the data used to populate a typical ballistics database. The graph demonstrates the relationships produced by the array of associations that can result in the database. It should be noted that in the case of Table 1 the only data used is details of the forensic ballistics items. This consists of examined bullets (B#), Cartridge Cases (C#) and Firearms (F#). Table 1 illustrates the results that humans can infer using current state of the art technology. Note should be made that Table 1 does not involve the use of a semi-automated ballistics system.

3. 
   Future State of the Art
   

Work we have undertaken in other areas where we have built extensive databases containing mixtures of data sets about populations of people, crimes and evidence items leads us to the conclusion that complex links, patterns and clusters can be observed given certain processing conditions. Leary (2004) demonstrated that in any data set where there are a range of variables and a series of inputs the potential for complexity will always exist. In such systems the ability to compare, distinguish, combine variables in an effort to reveal hidden relationships offers many advantages to knowledge generation. However, the problem that often arises is managing the ‘explosion’ of outcomes that often results. In any given series of variables there is the potential for a massive number of potential combinations of any 2 or more variables. The possible number of combinations from a data set involving any two or more items can be calculated on the basis of 2n-(n+1) possible combinations. For 10 the number of possible combinations is 1013, for 25 it is over 33.5 Million possible combinations. Therefore, even in a modest data set there are many potential avenues of enquiry for exploration and experimentation.

Implicit in these systems are self organising principles. These approaches involve the use of self organising emergent systems. These systems are self organising because the data processing conditions are set to run automatically and identify through several layers of potential association, hidden clusters of data that were previously beyond observation. These systems produce emergent properties because they process large volumes of data and iteratively search, locate and reveal relevant links and associations that could not be observed using conventional systems. Currently, these systems are not used in forensic ballistics. We are producing such a technology in 2009 to demonstrate the potential benefits of such an approach.

Table 2 and the Ballistics Matrix Key illustrate how broader conceptions of applicable and relevant data can be used to develop a richer conceptualisation of knowledge about crime. The graph illustrates how these additional items of data extend the knowledge revealed by the database. Currently this conceptual approach and technological functionality is not available.

T

able 2: Links between Firearms, Bullets, and Cartridge Cases using Additional Data about Crime

The future state of the art needs to embody a series of new concepts:-

1. 
   Broader use of data;
   
2. 
   Extending the types of data available;
   
3. 
   Data sharing between different countries;
   
4. 
   Self organising emergent systems (dynamic systems);
   
5. 
   Interoperable ballistics technologies;
   
6. 
   Pan-European Technology Network to facilitate data sharing, communications and data flow.
   

Interoperability of data and different systems present major strategic and tactical benefits. This is not currently in place at either a National level or International level. The benefit is gained by automating the routines of data sharing, correlations, processing and intelligent analytics on a continuous basis. Crimes can be routinely linked, threats can be monitored and situational awareness can be routinely managed with in-built ‘Alerts’.

4. 
   Conclusion
   

What is needed is a new approach and set of working principles and standards to enable the broader use of existing data and technology. Additional information that is routinely recorded about crime can yield additional links between crimes and the people that commit them if integrated relationally with technical ballistics data. Increasing the availability of this type of data and sharing it will also optimise current state of the art technology and approaches.

Current work involves researching and developing a system that:

1. 
   Utilizes ancillary crime data routinely collected by law enforcement organizations;
   
2. 
   Undertakes analysis of ballistics data and crime information that is systemized and semi-automated.
   
3. 
   Highlights the benefits of data sharing being adopted on a wide-scale;
   
4. 
   Improving and standardizing data capture techniques;
   
5. 
   Achieves technical interoperability of data and systems.
   

Bachrach, B. (2006). A Statistical Validation of the Individuality of Guns Using 3D Images of Bullets. National Institute of Justice Office of Justice Programs, U.S. Department of Justice. http://www.ncjrs.gov/pdffiles1/nij/grants/213674.pdf

Brinck, T. B. (2008). Comparing the Performance of IBIS and BulletTRAX-3D Technology Using Bullets Fired Through 10 Consecutively Rifled Barrels. Journal of Forensic Science. Volume 53 (3).

De Kinder, J. (2002). Ballistic Fingerprinting Databases. Science and Justice. Volume 42 (4), pgs 197 – 203.

Leary, R.M. (2004) Evidential Reasoning in Criminal Pre-Trial Fact Investigation. University College London.

Leon, F.P. (2006). Automated Comparison of Firearm Bullets. Forensic Science International. Volume 146.

Saks, M.J., and Koehler, J.J. (2005). The Coming Paradigm Shift in Forensic Identification Science. Science. Volume 309 (5^th August 2005).

Leon, F.P. (2005) 'Automated comparison of firearm bullets', Forensic Science International, Vol. 156, February, pp.40-50.