Nowadays many systems require features for recognition of video stream instead of static image processing. The core of these systems is a high-productive video stream recognition engine. This engine can be used in different applications:

1. Searching for video fragments in video stream
2. Searching and tracing predefined objects in video stream
3. Searching for texts in video stream and their recognition
4. Processing video stream for its optimization before transferring video stream through network
5. Management systems based on computer vision

All these tasks are rarely used in different communication, surveillance, video processing and security systems.
Problems of video stream processing and recognition in the reality are a complex of difficult tasks:

1. Video stream processing for removing doubled frames
2. Recognition of predefined objects (search and recognition)
3. Segmentation of video stream into similar fragments
4. Tracking of moving objects
5. Finding and recognition of optical characters
6. Minimization of processing time

We are working at the moment on the creation of a video stream processing and recognition platform. This platform will provide wide functionality for solving mentioned tasks and can be used in different applications.