|
|
|
AFIS, It’s Future (The following text is a paraphrastic transcript of the presentation at the October 1, 1994 SCAFO meeting)
Speaker JOSEPH BONINO I am going to try to excite you with some of the things that are going on in terms of the FBI's IAFIS and how it got developed and what's going on in terms of the enhancements to the CAL--ID system. I thought it would be helpful for those of you who don't have the entire background to give you a little history of how the FBI happened to get into the business of replacing its AFIS. I think it all started sometime in 1988 when Assistant Director Larry York invited about 50 people to the FBI for about 3 days to solicit user input. I think that was the first crack or chink in the armor where they began to realize that there was a lot of customer dissatisfaction out there and they had to stop telling themselves that they had the best system in the world. There were state and local agencies out there that were building new AFIS systems, who were really becoming experts in the AFIS world, and the FBI had a lot of catching up to do. In 1989 Director Sessions, who was really quite a mentor and believer in AFIS and new technology, came to an advisory policy board meeting and said, “I want a committee to come to Washington next week and I want them to give us a road map for how to get to the future and get us a new fingerprint system, so we can begin to serve our customers.” It wasn't quite one week later, but about three weeks later, a committee of seven of us that I chaired spent a whole week at the FBI building in Washington. We had an extensive look at what they do, brought a lot of our own user experience, (being most of us state and local AFIS users and managers) and we wrote a report called the Identification Division Revitalization which told the FBI really that it had to set a new course. Basically, they had to go to a minutiae based AFIS because what they had then and what they have now really is not a minutiae based AFIS. An internal audit report was done by the FBI's Office of program evaluations and audits, which was very critical about the thirty to forty--five days it was taking to get tenprint identifications done. The FBI realized that, unless they could come up with a system that could get people identified while they were in custody, using electronic modern technology and imaging, they really weren't providing service. A couple of the things that were in the report were that the new AFIS would be minutiae based and that it would be image driven too. There was a lot of emphasis by the FBI in the early days that they were going to somehow compel the different AFIS vendors to disclose their algorithms and therefore develop some kind of black box interface, but everybody knew that wasn't going to work. Only by having livescan developed and by having image transmission standards developed for telecommunicating fingerprints could you have inter AFIS operability. So that was a major operating requirement of the new system. Again, although you are primarily latent print people, it is kind of, a little difficult to realize for those of you who would like to have the same functionality at the federal level with respect to latent searching, that the FBI is building primarily a tenprint system. It will have a latent functionality and, because of the rate in which technology is expanding, we may get more than the minimum operating capability. The minimum capability is about 635 latent searches a day nationally and only half of those will be available to the states. There is a lot of hope, based on the kind of design competition that I'll get into later, that there actually will be a higher throughput than that, so that's a hope for all of you. Another key operating element was--—can you imagine going from thirty to forty--five days to a one day turn around maximum? There is a design concept “A day's work in a day's time.” If you can't get a day's work done in a day's time you never stop building a backlog. They are committed to get all their work done in one day. It will be a technological stretch to see whether or not they can meet an additional requirement they would like to meet, which is a two hour turn around for electronic transmissions of criminal arrest fingerprints. So while a person is in custody, assuming that your agency's requirements allow you to have that person in the lock up for two hours, you can get an identification. Now, there is a catch to that too. It's not two hours from the time you arrest them, but it's two hours from the time the FBI gets the fingerprints. So, hopefully, with some of the other things that I'll talk about, streamlining and store and forward, we can close that gap so that two hour period is meaningful to you and you really can get an actual identification while the person is still in custody. Again, it will have a latent capability that I talked about and actually one of the things I think a lot of state and local users imposed on the FBI was a rather rigorous performance capability for the latent capability. So, in terms of accuracy and through put, there was a very rather demanding requirement, in terms of 65% accuracy for those of you who are latent print people and know what that means in terms of AFIS benchmarking. Another key element in the requirements which set in motion a process which has culminated in the last few years is the establishment of National Standards--—standards for fingerprint card format, interoperability, resolution, livescan, and telecommunications. Now all that has been done. There were three NIST meetings held at the National Institute of Standards and Technology, they having published standards for all of those elements, and those had better be the bedrock for the design and implementation of IAFIS. IAFIS is going to be completely image driven. If you send them an electronic fingerprint card it will be processed essentially lights out as an image within the FBI IAFIS system. If you send them a card, they'll turn it into an image and process it as an image. So again, a very important design concept. From 1990 to 1992 there was an extensive period of studying, planning, preparation, RFP preparation, and a lot of dialogue between the FBI and vendors. There is a very elaborate federal acquisition process. I know each of you in your own jurisdiction probably have some pretty onerous procurement requirements, but at the federal level it's about times ten. They have to publish things in the Congress Business Daily, and answer endless questions from the vendor community. The reason I say this is to give you a little bit of an explanation, not a complete one, as to why it took from 1990 to 1994 to be where we are, where the FBI is actually awarding contracts, and are in a procurement and development stage. I guess many of you know this but, and I'll answer questions on it later, the basic building blocks of the IAFIS system are III, ITN, AFIS and FICO. III is a complete rebuilding of the criminal history driver for the system. The ITN, which we originally called Image Transmission Network, is going to handle all these images. It's now called Identification Tasking and Networking and the reason is that it's more than just image handling. It's workflow software, to handle images to the ITN and to the network manager for the entire project. It will talk to III and pass transactions back and forth from III. It will also talk to the AFIS circuit so it's a key work manager of the system and then, of course, there is the AFIS matching agent itself. The fourth element, which is critical, is what they call FICO, (Fingerprint Image Conversion Operation) which is the conversion and that conversion is an obviously monumental task when you're talking about converting over 30 million cards. In terms of scheduling, they have what is called the CJIS Master Schedule. It shows the major projects and where they are on a time line. It starts really with the 90--91--92 period where all the RFP's were prepared and it ends up with what they call FOC (full operational capability) of the entire AFIS in early 1998--—still three years away. But a lot of progress has been made. The FICO RFP has been out now for almost a year and they expect to award that contract in December. The III RFP was released in September `93 and the contract was awarded in August of this year (`94) to SAIC, Science Applications International Corporation. One of the benefits that the FBI accrued from having SAIC do the III is that they basically wrote the software for the original III system so they had an excellent operating knowledge of how the FBI criminal history system works, and would be in a good position, given that thorough understanding, to have a high probability of success in performing the III mission. The ITN RFP was released in April `93. The contract was awarded in April `94 to PRC Corporation, (Planning and Research Corporation) of McLean Virginia. The contract was about 75 million dollars and has a very robust architecture. It looks like it will really get the job done. No CPU, all de--centralized processing, very interesting architecture and very solid design. The AFIS matching agent for the system is one that's going to be able to give you 635 latent searches a day and 62,000 tenprint searches a day. The new matching agent will be done differently. There is a requirement in the Federal Government that when you can't prove that something will work by just putting out an RFP and awarding a contract, you have to conduct a design competition. Now they were reasonably sure that the ITN and III could be done with the kind of solutions that the vendors were proposing, but all of you know that the biggest AFIS around is the CAL--ID of California with about an 8 million card database. You're talking about a 30--35 million card database and through the life cycle maybe into the 40 million cards. They felt that they really had to get out there and do a design competition to see what the vendor community could do and they also had to do this for a risk management reason. They had to be sure they were going to get this job done. They had to push the technological envelope. There's a procedure called a circular A--109 which is used if they (Federal Government) want to build a fighter aircraft. They build two or three prototypes and fly them, hence the term “fly--off”. They released the design competition RFP for the AFIS in July `93. In May `94 they awarded three development contracts at approximately 10 million dollars apiece. The first was to Unysis who has NEC as a sub. The second was to TRW who has Cogent as a sub. The third was to Martin--Marrietta who has a working relationship with Morpho. They have a licensing inter--operable arrangement where they can use Morpho's fingerprint matching algorithms. The design competition will last through July 1995. There will be a four month evaluation period, then a final award and then they'll begin to build that system. All these things are shown on the time line together because they want to get them all to come together at the same point in time in 1998. Now, the ITN and III will be ready sooner. They'll be ready in the mid `96 time frame, so what the FBI is going to do is integrate those with their present IDAS system, (Identification Division Automated System). It's a system which needed fixing. What they will do is use it to do their AFIS matching, integrated with the new III and new ITN and provide some degree of better service until they can bring on the new AFIS matching engine in 1998 and integrate the whole system in Parksburg West Virginia and be able to provide the throughput services to which they have committed. They also had a very helpful system design review for the entire system a week and a half ago in Quantico and that was really a good thing for them to do. They only had about 30 users there. We faulted them considerably for that because, of the two hundred people in the room, about 50 percent were vendors, about 30 percent FBI personnel and about 20 percent were users. We said, “Fellows, you've got this backwards. It's the users who need to know what's going on out here,” and they listened. The point of that is that any of you who want to go on your own money, there are going to be four regional meetings. These SDR's are a two day review of the FBI IAFIS system, to gather user input and essentially market the system to the user. It should be in the first week of November, in Denver. When I get a little more information on that I will be glad to communicate it to your President and share that with you. If any of you, through your agencies, have the travel money or would like to participate in that, I think you'll find it very interesting. For those of us users who were in Quantico, the trip was worthwhile. I already had a fairly comfortable feeling about what they were doing because I am sort of semi--responsible for some of it. I think a lot of people who had misgivings saw what they were doing, saw how currently it has been designed, how well the risk had been managed, how thoroughly they had outside contractor help to insure that milestones were being met and systems integrations milestones were being met. I think a lot of people felt a lot better knowing what they were doing. There are a lot of documents associated with IAFIS and they have committed either to mail them to agencies or users who want them or to provide them on disk. So we can work with you again, for you who have an interest in that, we can get you the data you need. We're also urging the FBI do some more external marketing in terms of video tapes and Ombudsman, to get out and talk to people. I think the more you know the better you'll feel about it and the better you can plan for it. There is still some interest in issues that need to be resolved. One of the sub--systems or an option that was built into the III is an Interstate Photo System which is basically a mugshot system. The reason it was designed as an option was that there was a lot of pressure from Congress to keep the money down. Those of us who worked with the FBI to design this thing said, “Mugshots would really be nice, so let's design it as an option. Let's pattern it after CAL--Photo,” which you're all familiar with. If the economics and the technology are such that we can exercise that option, that would be a good thing. Well, it's almost a certainty now, given the award in III, that the IPS, because it's doable at a rather low cost, will be exercised. This raises a number of issues that again will involve you as users. I know Wally (Wally Briefs--Sunnyvale Dept. Public Safety) has been involved in some of these things. It was our suggestion recently at a NCIC meeting that the FBI work with the NIST to begin to invite the vendor community, the livescan community, the driver's license bureaus, the scientists and most importantly, the users together for as many conferences as it takes to work out what exactly are the requirements for a system. What do you want--—profiles? What level of resolution? How do you want them compressed? How do want them stored? How do you want to access? All of these are relevant issues and hopefully we'll get a lot of input through the NIST standards and processes and publish a standard which we can then hand to vendors, and say, “Do you want to sell us an imaging system?--—Here are the standards.” That worked very well in the fingerprint arena with live scan, and AFIS vendors to some extent, and it should work here too. Now there are also some issues related to NCIC. NCIC 2000 was on a development track starting in 1985. They built some fingerprint functionality into it, not knowing that the Congress was going to come rushing by them in about 1990 with 500 million dollars and a 200 million dollar building to build a national AFIS system or rebuild a national AFIS system. So to some extent, those requirements are in conflict. There are areas where many of the users, many of the largest users in the country, are NEC users and they use thumbs for tenprint work. Their (NCIC 2000) requirement was for index fingers. There are a lot of angry people or at least concerned people who are saying, “Why?” There are other things that possibly could be done in terms of using IAFIS as a source for file prints. Hopefully, thumbprints will be used in the wanted persons system in NCIC 2000. There is also a request from users, which the FBI is evaluating, to do a wanted persons system check by thumbprint at the time of a tenprint inquiry. You get a fingerprint based wanted person check. That, I think, would be a major improvement. So there again, they're beginning to look at that because they're going to begin to homogenize those systems. From a user perspective's view, I know LAPD is strapped for money and I think most of you are. NCIC 2000 is basically a 486 terminal with imaging capability. We don't know what the IAFIS access device is going to be, but we're saying to FBI, “Why don't you build it all into the one box, so you only have to buy the one terminal.” So they are looking very seriously at that, yet to be resolved. Since livescan is the hub of the IAFIS system, it is important to realize that a lot of progress has been made in terms of standard setting and so forth. The FBI also has addressed the whole issue of standards certification very creatively in the last year. They have what they call an engineering concept for equipment certification so that if a vendor, DBI, Identix or anybody else, wanted to sell to the law enforcement community an FBI approved livescan device, they had to go through an elaborate process with the FBI lab division to get the device certified. The FBI lab division was the first to say that they only have limited resources, but what this did was become a gating factor on the introduction of new equipment. Furthermore, suppose you're a vendor and you want to introduce a new model and you want to improve the scanner. Well, you'd have to go through the whole process again. So what the FBI has done is reduce all the major components of the livescan system to engineering requirements so if you're going to change one part, you only re--certify one part. That speeds the introduction of improvements into the system. I think that is progressing well, and they're actually so happy with that approach to equipment and technology, they're probably going to use it in terms of the NCIC equipment. An ominous cloud over livescan was raised by the FBI a few months ago and that is that they were operating on a live scan quality standard set in 1988. Subsequent to the development of IAFIS, they developed much more specific scientifically valid image capture standards which, for those of you who are latent examiners, I think was driven by the lab division who were very concerned about having the very best quality you could get to reduce it to its simplest elements. It's a set of specifications including modulation, transfer function contrasts, and none of the current systems technically meet what they call Appendix F (the Electronic Fingerprint Transmission Specification). So the FBI was talking about grandfathering equipment out there. They were originally talking grandfathering it for only a year. For those of you who are from LA county and just spent 5.8 million dollars, only having a grandfather clause for one year and then not having the FBI accept your cards sounds pretty serious. Well, we had a western region FBI meeting a couple of days ago, and we passed a motion which I think the FBI welcomes which is that they forget this idea about the limited grandfathering clause. There are only about 350 or 400 livescan machines installed throughout the country. We consider those as the installed base. To set the image quality standard for maybe mid--1995 and tell everybody that's when the FBI is going to require compliance with Appendix F, vendors will eat it. There are many people who have already spent money. City of Chicago has an extensive network, LA county does and there are several others. And so there's been progress, and because the FBI listens, at least in these cases, they have been particularly anxious to give customer satisfaction in this area. And again, those of you who attended the IAI Conference in Phoenix will know that a lot of progress was made on IAI taking a position on the compression algorithm the FBI was going to use. The so--called WSQ algorithm is set at a 15:1 compression ratio. There was also, very frankly, some confusion, I guess that's a nice way to put it, about exactly whether livescan was here or not, whether the FBI supported livescan, whether they were actually building the right system. Those of us in the policy--making positions prevailed upon the FBI to please speak with one voice. And they endeavored to do that after the IAI Conference. About two weeks ago the assistant director of the CJIS Division and the assistant director of the lab division got together, and I'll paraphrase the conversation in high--level terms. Basically the new director of the lab division asked rhetorically, “How far are we down the road here?” And Peter Higgins said, “Well, we've spent one hundred and 27 million dollars so far,” And he says, “Well, I think we're well on the way. Let's build this 500 DPI system and when technology allows us to upgrade, maybe to a full operational capability, we can upgrade then.” So this system is a 500 DPI system they are building. It's going to be operating in the WSQ algorithm with a 15:1 compression, and where we can introduce improvements on resolution or improvements in image capture, that will be done, but the FBI speaks with one voice and is committed to that system as designed and is moving forward. When Wally and I were in a conference in Quantico a couple of years ago, we talked about a number of issues related to how the functionalities of the latent system would be apportioned, how the unsolved latent data base was going to be structured, whether they were going to purge at a certain time. The unsolved latent data basis is going to be 250,000, but that would fill up in about a year and a half if you put everything in, so they are going to try and set some priorities. They are going to work closely with IAI, and I think you'll see a system regarding the administration of the unsolved latent data searches that will meet your needs. Again, I've told you about throughput, and I'll be glad to answer, at the end of the talk, any questions you have about the FBI system, but two final things. Just a little overview about what's going on in the CAL--ID and some new things that are in the works that you probably have heard about, but some you may not have heard about. CAL--ID is planning an upgrade. They are very likely getting the money next month to progress so that by 1996 they can begin acquiring an AFIS 21. They expect to expand their tenprint data base to about 12 million, their latent data base, permanent data base, to about 4 million. They probably will have two image verification archives separating tenprint and latent. The livescan store and forward original fingerprints will probably go into a gray scale optical file and it would be used for latent verification. So you'll have everything (resolution) they've got as an original electronic image to verify your latents against. It will maintain the DIRS (digital image retrieval system) images for tenprint verification. They are spending about $400,000 for store and forward. Store and forward is the way that all of the systems are going to operate in the future. Anybody who is capturing prints electronically is going to want it to compress them, probably search them once against the state AFIS, and then search them once again against the Federal AFIS. If there's no hit in the state system, the system stores it as a no hits, writes it to an archive file, sends it to the FBI, they do the same, an answer comes back by NCIC saying this is your FBI number or it's been registered as a no go. Store forward is also driving the design of the new California system. California has had seven or eight remote data bases in addition to the central site data base. It's a topic of lively discussion among the AFIS data base administrators in California as to whether or not they need to have a data base for tenprints. The answer is coming out something like this: Most of the agencies that have an AFIS system in California will probably keep them. They probably will not use them for tenprints. They probably will capture their booking prints by livescan. Their first search will be to the state through store and forward, the second to the FBI when they have access to the FBI, but they'll want to maintain their data base because they may have local offenses, they may have more complete records, and they'll do all their latent searches there. There is a truism in the forensic world that the best latent data base is the one that's local. You make most of your hits against local customers, so it's going to be a very hard sell for agencies to give up their AFIS's just because --— even because of costs since in Los Angeles alone we still make 95 percent of our latents on our own data base. So there are a lot of competing interests here, money, policies and so forth, but the tenprint world is going to be simplified greatly, and it will be driven largely by image, by store and forward and mostly through searches. Los Angeles county probably would do it that way, some of the others may do the same. WIN is not going to go away. It is going to be even more robust. WIN will probably be a switch for tenprint searches, and it will provide for the bulk of the latent searches for all the WIN states. We're also seeing a lot of development in the northeast and in the south and midwest developing essentially copycats of WIN, and those networks themselves may talk to each other. Now with an image standard developed by NIST and telecommunications protocol, it won't be necessary to send cards --— you could communicate by image with another network. You can communicate by image with a Printrac or Morpho state or county or city. You can communicate with WIN or anybody else by sending them an image. All of these things are becoming possible because of telecommunications standardization and a standard fingerprint image protocol. Something that is coming I think I mentioned before is a big drive toward criminal justice mug shot systems. The association of motor vehicle administrators is taking a major role in this. Possibly we'll see, in conjunction with the development of the smart card in the next few years, a microchip in a single smart card that people will have that will replace all your gas credit cards, all of your ATM cards, all of your credit cards, and, in addition, probably have your color photo and your fingerprint minutia, and that's not very far away. Also, I think based on the kind of technology we're looking at for the FBI IAFIS, even though it will be a stretch for the people that work for us in terms of the tenprint world, probably 60 to 80 percent of your tenprint identifications can be made automatically by the technology that exists right now. The only reason we verify all of these is people get nervous. Sooner or later we'll set an algorithm with a threshold score, and that's what they call a no false positive score, and you can do that and all the rest of them will simply be transparent. And the good part about that is a lot of people in our business, especially in the first generation of AFIS, were so terrified AFIS was going to take their job away. All this is going to do is redistribute the work. It's going to take the drudgery of tenprint work and set it aside. It's going to create more opportunities for people to be kind of intermediate level AFIS technicians. Our tenprint people will probably trace more cases and run more cases. They will be able to handle more of the fingerprint workload or the latent workload, and then the latent examiners themselves will have more court work and more file verification to make. There is plenty of work to go around. Technology is going to make us more productive as police agencies, and we have to remember that putting people in jail and preventing crimes and catching burglars and murderers --— that's what we're all here for, and this is what's becoming easier for us. Also the FBI held a technology conference last year. Two really interesting things came out. One is they are on the verge of or very close to developing a good matching algorithm for facial features just like people are developing matching algorithms for palm prints. They are very close to being able to not only having a mug shot which is kind of like a visual soft ID, it's called a soft ID, they'll actually have an algorithm that says this is a positive ID based on an algorithm that can match facial features. And that may ultimately be put into IPS later on. Also the FBI is talking about international links where we'll start linking to the British system and maybe the system in Germany, and we're linking with Interpol more. The FBI is working with Interpol more to share the stolen vehicle file. There has been a lot of internationalization of the criminal justice data base. And, finally, of particular interest to you is a new technology for capturing livescan or livescan fingerprints or palm prints and that is using ultrasound rather than light waves. So rather than the CCD image captured it uses sounds waves, ultrasound, and it looks like there's much less light scattering, there's a much higher level of resolution with very little noise and you get very clean detail and more detail and you can actually get subdermal features from the fingers. And it's a very promising research. This may be the upgrade to livescan in two or three years. Very exciting. It will create perhaps a whole new expertise in your forensic area. So with that, I'll be glad to answer any questions you have. (Editor--—Questions have been paraphrased) Question: Of the committees making recommendations, how many of the users have been latent print examiners? Answer: Well, the vehicle that they have had in the past to set the requirements for IAFIS at large has been system administrators and policy makers, but the FBI recognized that it needed to have the input of latent examiners. That's why they held several latent conferences. They held a three--day conference at Quantico in 1992 specifically to talk about user requirements, operational requirements, allocation of searches, the unsolved latent data base, and I think all during that time we had the expertise and constant advice and oversight, if you will, of the lab division of the FBI. Question: But are they (latent print examiners) going to be included in the future? Answer: Oh, of course, of course. This is a thorny issue, the issue of how much is enough for livescan, and there are many latent examiners who have their degree of concern. And what I don't want to do is convey what some latent examiners have tried to say which is “Those policy makers or assistant administrators, they really don't understand our needs.” I think we have developed--—because of that awareness of that problem--—we have developed a much closer link to IAI. Mike Fitzpatrick, the president of IAI, came to our identification services subcommittee meeting and has provided a close dialogue with IAI. We've got to bring the latent community along with us. We've got to resolve issues along with them. The advisory policy board of the identification services subcommittee that specifically deals with these issues essentially made a policy decision to move in that direction rather than try to impose a system design on the latent community. We felt that would be tragic. And if we didn't have the cooperation of the latent community, I think there would be a schism develop that ultimately would end up in some kind of unreliable --— a feud developing that would develop or show up in court testimony and we don't want that. So I think the lab division itself has been our closest watchdog. And I think in the area of image quality, the lab division has gone beyond just the latent examiners. They've actually had scientists involved in looking at the technology of resolution and so forth. They have had a project going on right now to be sure that as technology comes along, it will allow higher levels of resolution that we can afford, and I'm not parenthetically saying the issue came up in 1992 at the conference in Quantico. The FBI said we'd love to build you a 1000 DPI system but we need another billion and a half dollars to do it. I think that closed the question for a while. But still, that doesn't mean that we should be insensitive to the latent examiners. From a positive perspective, I'll give you my side of the story here. Our business is not a perfect business. We don't make a hundred percent of every case. We don't catch every crook. If you look at what the system will deliver when it's functional at full operational capability, you only get so much. What you miss on the margins would be insignificant compared to what you were able to do in the past and what you're able to do in the future. And I think that the normal anxieties and fears about the new technology and the fears about changing our standards are good things. And they keep us honest. At least you hope they do. Question: You are our spokesperson for this area, but I don't recall you ever reaching out directly to our local latent print community trying to get input from our perspective. Answer: Well, you know, maybe I didn't meet with SCAFO. I think there's been a lot of discussion with IAI, in general, number one. Number two, we've dealt with, we've gotten latent input from George Ito and other people from the state. You can't talk to everybody. I can't give you a perfect answer, but I'll tell you that the FBI has been widely criticized by people from all over the country who say, “You never asked us about NCIC 2000. You awful people.” Well, the fact was they sent out 1600 questionnaires to agencies of all sizes throughout the country. They sent out a team from Mitre Corporation and personally interviewed people from a hundred agencies in the country. They did a very thorough analysis of what the users wanted, but they couldn't talk to everybody, and a lot of people who weren't talked to felt that they were being neglected. I think these issues are being thoroughly and honestly evaluated and discussed with the latent user community. Question: If there would have been more input from the latent community, would the resolution have been higher? Answer: No. It's been limited by money and if you talk to the livescan people in terms of what it takes to capture a fingerprint with no unexplainable artifacts in them, we've crossed that bridge. People always want more resolution. The question is diminishing marginal utilities. How much more do you get if you double --— and you can't go from 500 dots per inch to 600. It doesn't work that way. You can go from 500 to a 1000, and we're not in that realm right now. So what are we going to do--—stop the train and wait? Question: At the FBI Symposium last year users from other countries talked about higher resolutions, i.e. 4000 DPI. Answer: There was a lot of misinformation and disinformation at that conference in Quantico. Nobody had their act together, least of all the FBI. That's been taken care of. I think they speak with one voice now. The laboratory division supports the systems being built. That system will get the job done and when technology and finances permit, it will be improved. There are a number of different ways you can look at the resolution requirement. The resolution requirement is two things. It's one for image capturing, scanning. It's another if you print the card. There are a lot of reasons that are leading us to look at capturing. That's why store and forward is so important. If you can capture that image and archive it electronically, you'd probably have all the scanner you'd need. I --— trust me. 500 DPI resolution will give you a card that is superior to most of the inked cards you get. The real question is training, equipment calibration, proper telecommunications, and video degradation. We need to be aware of those as latent examiners. You need to know what it is that you're dealing with as a product when you testify, but as far as unexplainable artifacts in the cards with a properly taken livescan, it generally is superior to an inked print, the average ink print. And there's an old maxim also that if we had tried to introduce ink prints right now, while we had been doing livescan for 80 years, it would be impossible. It just goes with the territory. You were there when we tried to bring in the AFIS. People thought the sky was falling when we were using digital images for verification on screen. People got used to it. Most of this disinformation came from the laboratory division and some particular people. These people are perfectionists. We don't live in a perfect world, but this system will get the job done and done well. Question: The FBI continues to come up with the latest and newest thing that they want done in order to integrate with their system. How many people are going to be on line if all they are going to do is to grandfather for one year? Answer: I just got through saying they are not going to worry about that part of it. What you have is going to be acceptable. And as a practical matter, the FBI will tell you that even if your equipment doesn't meet their standards, they are not going to reject your cards. Here's what was happening. I mean, I've tried to be discreet about this. The lab division basically was off in left field saying--—you know--—the sky is falling, the sky is falling. You cannot operate that way. The FBI had to speak with one voice. The fact was the people who were building that system would like not to have been pressured that way. This business about only a one year grandfather was the lab division's idea. We the users said forget it. They are now going to forget it. They are going to be okay with what you have. Question: What about the WSQ compressions standard? Answer: WSQ was a compressional algorithm for telecommunicating your fingerprint images. I suppose you could send raw images, in fact, you can send uncompressed images to the state if you want. The state will compress them and send them to the FBI. You don't have to implement the WSQ if you don't want to, but the cost you pay for not implementing WSQ (which is very reliable--—it has only three percent on the margins --— that might create some problems or you can compress it by tweaking it yourself) is time. Do you want a card to get to you in 2 minutes or 22 minutes. It's up to you. The FBI has to have a compression algorithm. They have to have a standard. When you buy a television you know that channel 3 is channel 3 because there are international standards that set those frequencies. They have to build a system that's going to accept an image when it comes in that it recognizes. That's what WSQ is about. It's about cost, and they are trying to help you and me. And that's not going to be a problem for you. Store and forward will handle that. Question: NEC is saying they may not continue to support the older systems. Answer: As far as NEC goes I mean, hell, I'm in that big time with our own system. They are telling us one more year of maintenance and then all bets are off. Well, I think we're going to negotiate that. And I think they want our business. We'll see. We've had unreasonable positions from NEC before. We spent $75,000 to develop some software which we were required to develop by contract so we could interface with LA County. And it didn't work very well. So we came in and said NEC fix it, damn it, it doesn't work. And they said $50,000, and I said no. I said I'm going to have Darryl Gates by letter give it to your big shot in Japan and he's going to tell him you guys don't support your big customer. The next day no more $50,000. Hopefully, we'll be successful with the maintenance issue. They're going to support this equipment. On the other hand, there is some finite point where you can't support computer equipment. It has a life cycle. That's one of the problems with law enforcement agencies. LA county has a personal history index. That's the criminal history system for LA county. It's been in operation for 18 years, costs them a million dollars a year to maintain. They are building a new one, but if things had been done properly with an appropriate life cycle for that system, it would have been dead and gone ten years ago. So there are life cycles to these systems and regrettably in the police information business the day you turn on your new system you better begin to plan the next one. Well, we've been trying to plan for a new AFIS system too, and it will be either--—maintain the old AFIS as long as we can for latents and then think about plan B--—or something else because I don't know where we're going to get another six or seven million dollars. But the trick is for us on the NEC issue, particularly AFIS 21, it's a lot more productive and relatively less expensive. I have 9 FMP's, they cost me $300,000 a piece in 1986. I don't know how much that's worth in 1994 dollars. But two of the new matchers will do what those 9 do and cost less than one of them. So I mean, I know we're all stretched, and we all want to do a good job, but at some point we have to find a way to get the money to get the new equipment. And the performance of the new equipment, I think, would be remarkably better. Question: I went from examining inked prints against latent prints to examining livescan against latent prints and I found it personally a lot harder or it took me a lot longer to do the livescan comparisons because they are not as clear as the ink prints. I figured that it would be me because I'm prejudiced since I'm accustomed to 17 years of looking at inked prints. But even new people were also indicating that it's harder to work with the livescan cards. You indicated earlier it's going to be updated in the future and that where it's going it will be better and better, then why don't we stop and wait until it does get better? You can't stop a train and make it wait, but you don't want it to wreck either. Answer: Well, you got several statements in there. Most of the problems with livescan images are in the operator training and attentiveness of the machine, and in some cases the telecommunications. There is nothing wrong with the device. The reason I know this is the FBI did a side--by--side test with 200 livescan produced cards and 200 very good ink cards, and of course the quality of control was done to get the livescan cards done correctly but with the currently available machines --— this was a year ago. They sent them out to about 15 different agencies and asked them to do comparisons, and to determine which was better, what the latent examiners liked the best. What they were going to do is they were going to tell people these are the ink prints and these are the livescan. We said “You don't have to tell them, just give them the prints. Put a number on them. Let them compare what they like. Let them see what the results are in terms of comparisons and makes and so forth, and then tell us which ones that they like.” By a small margin the identification rate was a little bit higher with the livescan, and there was a general tendency of the latent examiners to like the livescan prints more. So all I'm telling you is the things that you're seeing are problems --— and I talk everyday to Charlie Vannoy and Bernie Kammer (L.A. Sheriff's Dept. Records & Identification Managers). I know they recognize the problems that LA county has with livescan have to do with who is taking the livescan, not the machine. That's just a fact of life. But I will tell you I have seen some god awful ink prints too where people who were taking the ink prints don't really give a damn about what they are doing. So, you know, it doesn't mean that we have to stop the train--—because the technology is here--— but what we have to do is a different kind of training. We have to really make sure the machines are up to spec in terms of how they are tuned, and we've learned how to do that now. Checking the telecommunications and how they are printed. The printing is another problem. Store and forward will help us with livescan because we wouldn't have to print the cards. When you print the card --— if the card is captured at 500 DPI, you can only get 200 to 250 DPI in the print because of the problem of the printers. That problem is largely going away because there's a whole new generation of printers out there that perform beautifully at the 500 DPI level and that proves that we shouldn't have stopped the train just to wait for technology to catch up in some of these areas. But store and forward will allow us to deal with the images kept compressed, and with standards review we'll get all of the resolution, and I think you'll see a much better image to work with. I'm not trying to trivialize your concerns, they are fixed not by stopping the trains, but by getting people to run their machines better. There is one thing that the FBI recognizes. I mean people like me and some of them and others need to be on the stump a lot more. There's just been a lack of information. Maybe that makes your point, not that we made the wrong decisions, but we certainly didn't sell it well enough. I personally am the strongest critic of the FBI in that regard. I talked to Steve Pomerantz, the assistant director, two weeks ago and said, “You guys were going to come out with a video tape program, you were going to come out with an Ombudsman, you were going to come out with a whole lot more effort to sell and market your program and your system to the user community.” They realize they have to talk to you. Talking to people, getting information out, I think, will make a big difference in raising your level of confidence. Some of you are purists, fingerprint scientists--—you have a different view. I'm not going to change your view. I just hope you will be flexible enough to realize a lot has been done to try to meet you more than halfway. Question: More attention seemed to be focused after the latent print examiners stood up at the FBI Symposium in 1993 and insisted on being heard. Do you think that if the latent print community would have been involved from the beginning that there would have been a difference now? Answer: First off in terms of IAI, we've had a dialogue with IAI. IAI was present at the NIST conferences starting in 1990 where there was extensive discussion of latent images. Steve Meagher and Fred and I had a serious IAI discussion in front of 150 people on these subjects. I don't think it was 1992 when these discussion started to occur. The lab division has been vocal from the very beginning. IAI in truth had been quite concerned about this from the very beginning, concerned about livescan since about 1988. They really were opposed to livescan at that point. Mike Fitzpatrick, some of the others — there was a wide variety of opinions within IAI at that time. Some people agreed with us (I would like to call them progressive), and then there were people who were very critical. Meanwhile, technology was gradually improving. I think we could all see where technology was going. And at that time we were talking about 250 DPI livescan. Now it's 500, and it's going to be more. I know that I've spent countless hours at FBI headquarters talking with Steve Meagher and Danny Greathouse looking at these huge 14 X 14 blowups, looking at artifacts trying to decide whether or not their concerns were valid. We made a conscious decision we were going to court these people. Now, we could have done better. I mean, I have a job too. I can't go around the country speaking, but I think we could have done more and we pledge to do more. But I think all the technology is catching up fast, and I think IAI itself is getting to a much higher comfort level. I think they have come to the point where they realized that being part of the dialogue has helped integrate them into the development of this --— I think they are representing you very well. I have been in the trenches on this issue since 1987. There's been no shortage of input from the latent examiners from all over the place, the NEC user's conferences started in 1987. There's been a lot of discussion about a whole range of latent issues including digital imagery retrieval system and so on, but I'm sure we aren't perfect and far from it. I learned a lesson when we implemented our AFIS and that was as hard as I tried to reach out to the latent community and the tenprint people, we should have done about ten times more, because a lot of it was just mis--communication. I don't think that any of the latent examiners have any doubts about the AFIS or the digital images or anything else, but they had a long period of discomfort because of insufficient dialogue. And I think that there's a parallel and I appreciate your concern, and I understand your sensitivity because of your profession and when you testify as an expert, but hat's the exciting part about all of what's happened. Your level of expertise is going to go up. Happily the number of crooks we're going to catch with all these new tricks is just going to quadruple and, by working together and talking to each other, we'll find a way to get through this. (Editor--—In Mr. Bonino's presentation there was a wealth of detail about the massive IAFIS project. Some of the information is extremely enlightening and some thought--provoking. A consistent theme prevails—more communication will provide better understanding. Let's continue to the expand the lines of communication. SCAFO members and readers of THE PRINT want to hear your thoughts on these matters.)
This article was originally published in “THE PRINT” 10(9), October 1994,
pp 1-12 |