|
One of the main VIT Ltd. activities is the development of video surveillance systems (turnkey solutions), number plate recognition (Overseer AutoCode), face recognition (Overseer FaceCode), van number recognition modules, and video-capture boards. Overseer AutoCodeAutomatic number plate recognition system Overseer AutoCode is designed for searching, revealing and recognizing state number plates. Overseer AutoCode creates a chronological archive of number plates that can be additionally viewed and processed by an operator. The archive contains information about recognition time, number plate (text field), country, moving direction, relative speed, cut out frame with number plate, and the whole frame with car, parameters of which (color, mark) can be identified by an operator. Features: - recognition of several number plates in the camera’s view in case if the snapshot number plate recognition is possible;
- using different number patterns (e.g. number patterns of different counties);
- adding new number patterns;
- moving direction recognition;
- defining the relative speed of the vehicle (calculated depending on the camera’s technical features);
- connecting a movement detector to reduce the calculation needs to identify the number plate;
- setting separate detection areas;
- setting minimum and maximum expected number dimensions.
Overseer FaceCode Overseer FaceCode face recognition system is a software complex, designed for automatic face recognition and saving the information to the database. Intelligent hart of the system is based on applying the latest neural network algorithms. This provides high recognition level. Face recognition Overseer FaceCode system provides automatic image processing, search, capture, and comparing with the database for its identification. Overseer FaceCode implements two functions: face detection function that helps creating a chronological archive of faces, highlighted from the general background, and recognition function itself. Technical characteristics:
(Provided technical characteristics are based on the results of recognition algorithm testing with use of BioID (Switzerland) and FERET (USA) face databases) - time of face searching in the frame: < 40 ms;
- correct face capture coefficient – over 95% (BioID);
- time of face identification by comparing with 1000 records from database – under 50 ms (FERET);
- correct identification coefficient - over 95% (FERET);
- probability of mistaken access granting (FAR) — 1%;
- probability of mistaken access denial (FRR) — 4 %.
Board development: read more...
|
Video capture boards for PCI and USB interface were developed for video information input to the computer and further data processing.
The board is series-produced with the combined positive method on material FR-4. This completely meets requirements of the international standard DIN EN 60249-2.