DELTA-S is a fully functional Automated Fingerprint Identification System. It provides automated tenprint-to-tenprint search and latent-to-tenprint search. The system also has a full range of latent-and-tenprint enhancing capabilities. It is in use internationally by local, state, and federal law enforcement agencies. DELTA-S operates on varies UNIX platforms.

AFIS Open Systems Architecture

The Cogent Systems AFIS product line features a flexible, modular, software-based solution within an open systems architecture. Our AFIS products execute on industry standard UNIX platforms and benefit from the price/performance improvements in the marketplace.

New features and enhancements to our AFIS product are made available in the form of easy-to-implement, straight forward software upgrades. Cogent's AFIS solution takes advantage of the cost competitive UNIX workstation market which, we believe, could double the performance of the hardware platforms on the average of every two and a half years, while maintaining the same competitive cost. The software-based design ensures that the Cogent's AFIS solution will accommodate newer commercial hardware technology and protect our customers investment for years to come.

AFIS Accuracy

The accuracy of the Cogent algorithms has been benchmarked by numerous customers and has been proven to be one of the most accurate AFIS systems available today. The Cogent algorithms currently support operational sites ranging in size from 20 thousand records to 4.5 million. Automated quality assessments of a search print or latent dynamically determine the processing path of a search transaction to consistently achieve higher matching accuracy rates.

AFIS Reliability

Our AFIS system will incorporate fail-safe techniques and features to meet the requirements of our customers. These techniques include redundant servers that Automatically re-configure themselves in case of failure; redundant networking subsystems that protect against inter-system communication failure; disk mirroring to eliminate the possibility of lost data due to media failure; and a system of Automated data base backups to ensure that key data is kept up-to-date in the event of a power failure.

AFIS Automation

Some tasks a computer does well, others a human being can do better. The Cogent AFIS solution is designed to combine the best of automated and manual AFIS processes. Routine tasks are handled by the computer automatically, but the system is designed to request human intervention whenever necessary and appropriate.

For example, after a fingerprint card is scanned, the system automatically performs a series of image processing and data management routines to locate individual fingers, crop and clip images, identify minutiae and core locations, and determine fingerprint pattern classifications. These steps illustrate the difference between operations that are appropriately machine-controlled and those that require human intervention.

Cropping the images requires no human intervention and is, therefore, a completely automated process. Minutiae and core location operations are automated, but a machine assigned quality score determines whether subsequent human attention is required and, if so, the system automatically alerts an operator. Likewise, fingerprint classification is determined automatically, but, due to its critical nature, a subsequent review by a human operator is fully supported in order to allow a trained examiner to override the system's assigned pattern selection whenever necessary.

In the latter case, the system automatically references both the machine and the operator-assigned pattern classifications, reducing the possibility of missed matches due to miss-classification. As these examples indicate, the Cogent AFIS solution applies the power of computers to assist, not replace, the human fingerprint expert.

AFIS Functionality and User Productivity Tools

Cogent has incorporated a full set of useful tools, and the associated functionality, to assist fingerprint examiners to be more accurate and productive. Tools such as automated minutiae and core detection and automated pattern classification are standard features. These tools are supplemented with a powerful automated quality assessment feature that alerts the operator whenever human intervention is required to edit poor quality data. Other image manipulation features provide the examiner with easy-to-use tools to compare images, verify or deny positive identifications, and print reports that include images, annotations and text.

AFIS Ten-Print Functionality

Ten-print functionality includes the capability to search ten-print cards against new ten-print cards and ten-print cards against unsolved-latents. The system fully supports the scanning of various formats of ten-print cards. The capturing of associated demographic data can be performed by the examiner at the same time as image capture, or separately at a later time by clerical personnel. After capturing the fingerprint images the system automatically crops, clips and processes the ten-finger images. Automated quality assessment scores are created. Plain impressions can be automatically scanned and viewed to perform sequence checking against rolled impressions. The system fully supports the replacement of rolled impressions with plain impression for those instances where only poor quality rolled impressions are available or rolled impressions have been mistakenly entered into the wrong locations on the ten-print card. Standard tools available to the examiner for ten-print processing include the following:

• Automated minutiae and core detection
  
• Automated ridge pattern classification
  
• Automated ridge direction determination
  
• Automated minutiae type determination
  
• Automated minutiae quality color assignment
  
• Automated capture of logical rolled print area
  
• Automated capture of plain print area
  
• Visual comparison of rolled prints vs. plain prints
  
• Automated replacement of a rolled print by plan print
  
• Manual editing of minutiae and core locations/directions
  
• Automated pattern classification and secondary pattern referencing