Viinex Foundation 1.4 is focused on providing its users with access to a set of video analytics functionality. These are vehicle license plate recognition engines, railcar ID number recognition engines, and a number of traditional detectors: motion/activity, image quality, camera shift/tamper, - present in nearly every video management system.
License plate recognition
Viinex Foundation 1.4 has support for license plate recognition engines from several vendors. The availability of several integrated engines provides a choice to obtain high-quality results under specific site conditions (that is, prevailing license plates of a specific country, the vehicle speed at the moment of car registration, availability of CPU resources, etc). Viinex Foundation 1.4 provides a client application with a unified API to acquire the license plate recognition results so that no source code changes to the integrating software are required in order to switch to another LPR engine.
There are two major modes implemented for LPR in Viinex Foundation: a free pass mode and checkpoint mode. In both cases, a video stream is processed, - this makes it possible for client software to not know when the car appears within the sight of a camera, and use Viinex 1.4 as a car detector.
Viinex Foundation reports the following data to the client application:
- recognized license plate number;
- state affiliation of a vehicle;
- expected accuracy of recognition;
- coordinates of the license plate image in the frame.
Based on the received timestamp, the external application may acquire from the video archive an image or a video fragment containing the view of the respective vehicle.
Viinex Foundation 1.4 implements a consolidation module, which aggregates license plate recognition results from multiple cameras installed at the vehicle passage point, and then transmits the consolidated data to the integrating application. The latter handles recognition results and media data from several cameras - data related to a single vehicle pass - as a whole. Even more, the consolidation module makes it possible to significantly increase the rate of correctly recognized license plates by means of combining the recognition results of LPR engines from different manufacturers. These engines are based on different mathematical principles and methods, and therefore have different causes of recognition errors. When combining the recognition results, statistical independence of these errors makes it possible to obtain higher reliability of the consolidated result than the reliability of the result of each algorithm separately.
Railcar number recognition
For more than 10 years we've been developing and shipping the software for railcar identification. There can be numerous issues in integration with third-party equipment, software and end-users' business processes, of which we are aware. Viinex Foundation 1.4 supports several modes for railcar number recognition, railcars separation, and provides flexible means for gathering recognition data. In addition, there are some unique features in Viinex Foundation 1.4, like fisheye lens compensation for sites where videocamera is installed closely to the railroad, train maneuvering support, and the module for optical fluid level estimation working on images from IR cameras.
Viinex Foundation 1.4 is successfully deployed on many sites in ex-USSR where railcar ID number recognition is required for rolling stock identification and registration.
Motion/activity detector as well as image quality, camera tamer and camera shift detectors became a must-have features for PC-based video management system, and they can be used in applications for ATMs.