WINPHO Overview

What's the Nature of the Mugshot Problem?
Modern digitized mugshot systems are a tremendous asset to law enforcement. However, their benefit remains highly localized. Criminals, a naturally mobile sector of society, benefit from the relatively static nature of the mugshots they leave in their wake. While digitized mugshots tend to accumulate in databases well-suited to remote retrieval, those databases remain, for the most part, discrete and insular islands of information. It remains the case that for the vast majority of law enforcement agencies, the best way to get a high-quality copy of a mugshot captured by another agency hundreds of miles away is through the mail. So, although a police officer in Cheyenne, Wyoming can, within two minutes of forming the idea in his mind, be looking at the fingerprints of a person who was arrested in Carson City, Nevada, he will wait days for the mugshot that was captured at the same time as the fingerprints. Such is the case even though the mugshot image exists in an electronic file 1/20 the size of the fingerprint image, and is infinitely more useful in securing witness identification. This state of affairs is a boon to criminals and, given the otherwise high quality of the images and ease of local access, a tragedy for the criminal justice system.

Not a New Problem [i]
In 1870, as part of what we'd now call a "proof of concept" project, Eugène Appert was directed to photograph prisoners of the French military, beginning with some 100 in the prisons of Versailles. In January, 1871, Appert, as part of this project, photographed seventeen year old Henri Joseph Paloonnier. Met with praise, Appert's project went "production," and the photographing of prisoners became a regular part of their incarceration. Unfortunately, as is the sometimes the case with data-warehousing projects, the volume of data quickly outstripped the users' ability to organize them. In 1873, only three years after the first photograph was captured, the Préfecture de Paris found that it could no longer locate Paloonnier or any other specific photograph among the more than 100,000 it had accumulated. Lacking a systematic process of cataloging the images, it had become impossible to connect offenders with the photographs that could have identified them. It would be twelve more years before the Préfecture finally regained control of their records system and was able to make mugshots available to law enforcement. The result of this long, expensive exercise was, however, well worth the expense. In fact, it was a watershed event, resonating far beyond the 19th Century and Paris' twenty districts After 1885, thanks to the system of Prefect Alphonse Bertillon—a classification scheme immodestly called "Bertillonage"—police officers, prosecutors, and jailers were able to walk into the Préfecture records office, request a mugshot from the collection, and have it brought to the counter by a records clerk.

Today, in the vast majority of jurisdictions in America, a police officer, prosecutor, or jailer in need of a mugshot acquires it in exactly the same way that her 19th Century French counterpart did: she travels to the appropriate criminal history repository, walks up to the counter and asks for the photograph, whereupon a records clerk locates the requested image and provides it to her.

This is not to say that the business of capturing, storing, and locally retrieving mugshots has not progressed. On the contrary, today's mugshot systems are often technological marvels. But the problem of disseminating the images stored within those marvels of technology has remained unsolved since one of Bertillion's clerks first handed a photograph to a policeman in 1885.

This problem is not especially acute when booking photographs are used within the agency that captured them. Modern mugshot systems often provide for access from multiple terminals within the same agency, so long as the appropriate license and equipment are purchased. What's more, some commercial vendors have even begun to provide interfaces between the systems of different departments, although they do impose one very important condition: these systems can only be interfaced to other systems purchased from the same vendor. This has meant that, if a city like Hillsboro, Oregon, located some 30 miles west of Portland, wants access to the digital mugshot database of the Portland Police Bureau, it must submit to the expense of a system purchased from the same vendor that services Portland. So, while it must be said that the dissemination issue is being addressed by commercial vendors, those vendors are doing so in only the most parochial manner.

Hillsboro, as it happens, is an excellent subject for the study of this issue. Feeling the influence of crime in nearby Portland, Hillsboro officers find themselves in constant need of mugshots from the Portland Police Bureau database. However, until recently, acquiring those images has been a matter of sending a detective to Portland to physically bring a photograph back—a task that would take the better part of a morning. And while Hillsboro ostensibly had the option of purchasing a digital mugshot system from Portland's vendor, as practical matter it could not justify such an expenditure.

Hillsboro's problem is hardly unique. And the more geographically removed a police officer is from the mug shot he needs, the more acute the problem becomes. Although the rise of digitized booking photo systems would seem to have brought with it the promise of a solution to the question of how best to move these images over long distances, it has not. And the methods by which law enforcement agencies have historically worked around this problem have been stop-gap, at best.

When approached with the notion of providing broader access to data housed in the systems they sell, commercial mugshot system vendors traditionally object on two grounds:

• "If we allow agencies that don't have any system at all to interface to one of our systems, there will never be incentive for that agency to purchase a digitized mugshot system at all. This will reduce the size of our potential market."

• "If we allow agencies that have purchased a mugshot system from one of our competitors to interface with a system that we built and sold, it will tend to make our system less unique and erode our market share."

Given these objections, vendors have been steadfast in their refusal to accommodate a solution to this problem. And because these systems store data in a proprietary format unique to a particular manufacturer, it has been all but impossible for users to build their own interfaces, even to their own data. It is not uncommon for data to be stored in such a way so that both the collections of mugshot images and descriptive data can be located in the database, but no means of associating a particular image and its descriptive data is provided. Rather, vendors closely guard the algorithms used to put the right photo with the right text. While acknowledging that the data themselves remain the property of the agency that captured them, these vendors still call all the shots. The alternative is unthinkable and would put the law enforcement agency in the position of the Préfecture de Paris in 1873: in possession of thousands of photographs but unable to identify them.

A Solution
Mugshots have proven themselves indispensable to law enforcement. Simply put, if there is such a thing as criminal recidivism, it would be madness not to photograph every criminal who ever found himself in police custody. And if criminals tend to operate without regard for administrative boundaries such as county lines, city limits, and states, it would be equally mad not to have access to photographs of criminals that were captured by other law enforcement agencies. While there obviously was a time when sharing these photographs was, to say the least, impractical, such is no longer the case. Today, an agency that continues to use file cabinets to store mugshots has no notion of the value of a digitized system. An agency that uses a digitized system but does not network its images does not fully appreciate the incredible value of a mugshot. And an agency that does not photograph the individuals booked into its facility does all law enforcement a disservice. As has been shown to be the case with fingerprints, the true value of mugshots is not in capturing them, but rather, it is in making them economically, immediately, and seamlessly available. To skillfully capture the likeness of a criminal and make that image available to the largest possible sector of the criminal justice community is to confer a disproportionately high benefit.

WINPHO was designed to directly address the issue of sharing mugshots by showing three assumptions to be valid:

1. A standardized, rational, low-cost networking approach can be applied, using, to a large extent, existing infrastructure, to bring together the dissimilar mugshot systems in use today, thus leveraging the individual databases of many WIN members;
2. Adherence to a standardized scheme makes participation in this type of project much more affordable than has been portrayed in the past;
3. This kind of system is incredibly useful.

WINPHO's use of off-the-shelf hardware and a standard web browser user interface make this approach desirable from both fiscal and operational points of view. An agency may purchase a full-featured, commercial mugshot system and interface to WINPHO, buy a basic mugshot capture system with a native WINPHO interface, or simply put an existing desktop PC to work as a WINPHO receive station. In all of these cases, the connection to WINPHO is probably the least costly aspect of implementation.

Cost of Participation
WINPHO is priced on a cost-recovery model. WIN, as a not-for-profit corporation, ensures that all benefits of this program inure to its members. This means that, while a portion of WINPHO operating expenses are derived from the WIN general pool of resources, others are billed directly to the user. This structure is designed to provide WIN members with the most cost-effective approach possible in a stable, predictable pricing structure. Given the extensive use of existing WIN networks for transmission of WINPHO data, as well as the tremendous opportunity for users to employ hardware, software, and techniques they already possess, there is virtually no financial barrier to participation in WINPHO. In the rural jurisdictions that make up most of WIN, this is an enormous benefit.

Rational Approach
"Standardized" and "low-cost" are not viable arguments for a system that doesn't make sense to users. Those ideas do, however, inform the argument. To make WINPHO work, WIN had to present a system that, above all, appealed to police officers as a useful tool—it had to be "cop friendly." So, we talked to cops. The resulting product was built using carefully vetted criteria, and some bells and whistles were purposely left out.

WINPHO is a system that does a number of things very well. Fundamentally, it allows agencies to share mugshots. It is, at the same time, a "Rosetta Stone," translating data from different systems into a common format. Further, WINPHO builds photo lineups from the mugshots selected by the user. What's more, can be a research tool, providing, through log analysis, an understanding of the kinds of information that criminal investigators seek and the sources they use when they look for it. The ability to compare where a detective looks for clues and where he finds them could provide some revealing data. What WINPHO is not, however, is a high-end mugshot capture system. While mugshot capture is a basic function of WINPHO and available to all users at no additional charge [ii] , it has never been WINPHO's intent to compete with commercial mugshot system vendors. Our primary goal has been and continues to be the pooling of data through the development of interfaces to commercial systems—this done with the cooperation of those commercial vendors. Put another way, whether users opt for basic WINPHO-native systems or high-end commercial systems, the fundamental purpose of WINPHO is to aggregate all available, high quality mugshots, regardless of origin, and make them available to the broadest possible law enforcement audience. To this end, we believe that WINPHO interoperability adds value to every interfaced commercial system. While this approach may obviate the marketing argument that says that an entire state or regional network must be serviced by a single vendor, removing the single-vendor requirement makes each potential user within that market a more viable customer. Interestingly, this arrangement may result in greater market penetration, the effect of which would be more systems being sold and more mugshots being shared. When a vendor is chosen at a state level, any agency that wishes to participate must make that vendor's product fit that agency's operation. If the agency cannot afford the system or has operational objections, the only alternatives are 1) to contract for the development of a custom interface to the system that the agency can afford and finds to be an operational fit—an undertaking that can turn out to be even more expensive, or 2) forego the capture of mugshots. However, if the state establishes an open-architecture system to serve all users, and those users write RFPs requiring that vendors include an interface to that state system (a much more cost effective endeavor [iii] ), the universe of potential vendors expands. More affluent agencies may still acquire spectacular, cutting-edge mugshot capture systems, but doing so will not foreclose the possibility of less affluent agencies finding something in their price range that will work with the statewide network. In short, networked mugshot systems need an open-source server at their core. Any other alternative will serve particular vendors without serving the real purpose of the network: to share all mugshots among all users. Open systems serve the users first. Overall, the result is better coverage and a more useful database. This approach can be applied even in those jurisdictions where statewide or regional systems have already been procured, allowing agencies whose participation had been foreclosed by adoption of a single vendor access to the system.

WINPHO is, above all, a rational system. It just makes sense. Acquiring, storing, retrieving, and displaying mugshots is intuitive and fundamental. Using WINPHO for all mugshot-related tasks results in a work product that is useful and conforms to all applicable standards. However, simply using WINPHO as a conduit to share mugshots captured by a commercial system also leverages this resource. Elemental and pragmatic, open and scalable, this concept should be at the heart of all mugshot systems.

Use of Existing Infrastructure
WIN members are networked thorough their own state backbones and WIN/NLETS, the WIN-administered portion of the National Law Enforcement Telecommunications System. Transporting mugshots and fingerprints over this secure, private network is an ideal, but not unique solution. Other states and regions enjoy a similar infrastructure. Because this system was designed with an eye toward the transmission of fingerprint images, it is a natural fit for mugshots. This is due to a number of factors: First, the average mugshot record is roughly 5% the size of the average full AFIS record. Second, these networks are designed with excess bandwidth, with sometimes as much as 50% of the network bandwidth remaining available during an average period. Third, AFIS traffic has dramatic peaks and lulls, with some periods showing almost no activity at all. WINPHO has two different types of impact on the network. The most obvious impact is in the request made of the system by a user in search of a mugshot, and the fulfillment of that request by WINPHO. However, the effect of these transactions on network traffic is less than would be expected, given that 20 mugshot transactions use the same amount of bandwidth as one AFIS transaction. Rather, the more substantial impact on the network comes when scheduled uploads from WINPHO's capture stations occur. During these uploads, thousands of records may be sent to the server by each capture station. Thankfully, these uploads are unattended and can be scheduled any time, day or night. By evaluating AFIS traffic and scheduling mugshot uploads during AFIS lulls, the most efficient use is made of network bandwidth. This approach also applies to the daily WINPHO server updates of state proxy servers.

[i] Related areas such as facial recognition, synthesized mugshots, and the mugshot feature search problem were not addressed in this study. However, WINPHO's standardized, open architecture stands ready to support these features as they are integrated into commercial systems.

[ii] "Mugshot capture is available to all WINPHO users at no additional charge" means that this function is a fundamental feature of the software that WIN developed under this grant. We hasten to point out that additional hardware, such as a camera, is required to make use of this feature, and that the user is responsible for procurement. Also note that the monthly support charge for capture stations is $32.00, while receive stations are billed at $8.50 per month per workstation.

[iii] It is important to understand that all interfaces are not alike. Even among vendors who claim NIST-compliance, the mechanics of writing an interface to mugshot data is a complex undertaking. This is because vendors' claims to having a NIST system often refer only to the format of the image, leaving storage and transmission in a proprietary (non-NIST) format. While vendors can easily build access to their own data, the proprietary nature of most systems leaves them unable to build interfaces to data stored upon other systems without considerable investment of time and money, not to mention cooperation from the builder of the targeted system. However, most all vendors, and especially those who are NIST-compliant, can cheaply and easily interface their own system to an open, true NIST database.