TollPlus

header_case_studies

Tampa Hillsborough Expressway Authority (THEA) & Miami Dade Expressway Authority (MDX)

Tampa Hillsborough Expressway Authority (THEA) & Miami Dade Expressway Authority (MDX) logos

Situation

Both the Tampa Hillsborough Expressway Authority (THEA) and Miami Dade Expressway Authority (MDX) were looking for ways to enhance their image review process and resulting accuracy.

Solution

TollPlus developed and implemented an innovative web-based image processing application to significantly reduce manual image review and operational costs. The new product, called Enhanced Image Processing (EIP) at the time but since changed to ToP-IPS, was launched in 2013 at the Tampa Hillsborough Expressway Authority (THEA). The system processed customer license plate images whenever electronic (RFID) transaction information was not available, and additionally provided:

  • Integration with a third-party OCR
  • Integration with a third-party fingerprinting software
  • Integration with the host lane system
  • Service Oriented Architecture (SOA)
  • Java-based lightweight transaction processing engine to manage high volumes
  • Single sign-on
  • Configurable business rules
  • Clustering and load balancing

The system was deployed in May 2014 at the Miami Dade Expressway Authority (MDX) and processed over 450,000 transactions per day (1,500,000 images). The TollPlus ToP-IPS was developed with configurable and intelligent business rules to reduce the manual image review by introducing double-blind reviews to eliminate human typos, error traps, and DMV information, and minimize easily confused character read errors. Innovative accuracy enhancements included Probe Transactions to monitor human inattentiveness and grouping of same-vehicle transactions to reduce manual review of low confidence ALPR results.

Benefits

  • Performance

    After less than two months of operation at THEA, the system produced an audited read accuracy of more than 99.95% for the full range of license plate images. This performance was unmatched by any other system in the U.S. at that time. The system processed approximately 35,000 transactions per day (180,000 images), which involved processing the very dense payload camera images from the roadside. Delivery was provided on time and there was zero downtime throughout the contract term after the system went live. A 100% billing accuracy rate combined with a 40% reduction in operating cost were both realized.

    The system produced an audited read accuracy of more than 99.95% for the full range of license plate images at MDX. It also provided for faster revenue recognition by the agency along with a halt to revenue leakages.