Page 112 - Automated Fingerprint Identification Systems (AFIS)
P. 112

FROM PRINT TO IDENTIFICATION       97



          searches while also providing a backup should one string have to be shut down
          for maintenance, testing, or upgrade.
            Tenprint to unsolved latent (TP/UL) matchers contain the image charac-
          teristics of latent print images that were searched on AFIS, but for which no
          identification was made. Depending on the needs of the identification agency,
          these matchers can be configured to search some or all of the tenprint records
          that are inserted into the database each day. This search may require slightly
          more time than a tenprint to tenprint (TP/TP) search. In TP/TP searches,
          only the index fingers (or thumbs or both) are used. Unlike latent prints, these
          tenprints are usually of good quality. In TP/UL searches, however, the image
          characteristics from all ten fingers of a new tenprint record may be searched
          against the entire database of latent print images, some of which will be of poor
          quality. Agency administrators can determine the number of searches allowed
          and the capacity of the unsolved latent file. Control of the file can also be
          achieved through periodic purging of records, for example, when the statute
          of limitations for the crime from which the latent print was retrieved ends. If
          not purged, these cases may drain resources that are better spent on active
          cases.
            In the latent to tenprint (LT/TPlc) search, the image characteristics from
          the latent print are searched against the entire latent cognizant (or latent cog)
          database. This database, which contains the image characteristics of all ten
          fingers, may be a subset of the tenprint database or may be identical to the ten-
          print database, and it may contain more image records than the tenprint data-
          base. For example, if there are five million tenprint records on the tenprint
          database, then there are ten million image records (assuming the system uses
          the two index fingers) on the tenprint matchers. But if there is a three million
          record latent cog subset of these records with ten images for each record, then
          there are 30 million image records (three million records ¥ ten fingers) on the
          latent cog matchers.
            In addition to the two-finger tenprint matchers, the ten-finger latent cog
          matchers, and the single-finger unsolved latent matchers, there may also be
          matchers for palm prints (see Fig. 5.6). Palm images require a large amount of
          storage and transmission bandwidth. The AFIS configuration in Figure 5.1
          shows palm print searches and storage with similar functionality to the tenprint
          storage and searches. AFIS may be configured to include a matcher string that
          can search palms against other palms in various ways. This function is of par-
          ticular interest to latent print examiners, since palm impressions found at crime
          scenes can only be searched against other palm images. If there is database of
          known palm images to search against, the chances of making an identification
          from the evidence left at a crime scene greatly improves.
   107   108   109   110   111   112   113   114   115   116   117