SeeStorm Computer Vision Technologies

SeeStorm Computer Vision (CV) Laboratory has 15+ years of experience in image processing and machine vision. SeeStorm delivers CV algorithms both for real-time and non real-time image processing.

Face Feature Extraction technology (FFE) provides identification of all significant face features, such as eyebrows, eyes, lips, nose, etc. Face Feature Extraction works both in real-time mode with a video stream, and with static 2D images.

Face Feature Extraction output data (a special feature state vector) can be used in different high-level technologies. For example, Face Feature Extraction is an underlying module for Face Recognition, Mimic Recognition, Head Modeling and many other techniques.

SeeStorm Face Feature Extraction technology is available as FFE SDK, a stand alone software ready for integration into target applications.

Head Tracking technology determines the spatial position of a human head as a 3D object and tracks changes of its position and turning angles in real-time with high reliability.

This technology employs a set of biometric methods based on analysis of a video stream.
Head tracking technology in combination with Face Feature Extraction technology makes it possible to display the full range of a human head's motions on 3D models: the head of a 3D character (avatar) copies the user head's motions, with natural-looking nodding and tilting. This unmatched freedom of movement is one of the features available only from SeeStorm.

Mimic Recognition determines facial feature's motions and then identifies the person's mimic state. This technology is used for avatar (character) animation, expression tracking, phoneme synchronization, powering avatar chat communication via the Internet, etc.

Based on Head Tracking and Face Feature Extraction technologies, Mimics Recognition technology is robust against a wide range of lighting and background constraints.

Face Recognition technology is dedicated to recognizing (identifying or verifying) an authorized person even if the camera type or angle of view is different, or if the person has a new haircut, glasses, or a different facial expression.

Face Recognition allows locating a human face in a series of digital video images and then uses biometric techniques to define the individual facial features in terms of geometrical data.

Each personal data state vector is stored in a dedicated data bank. Original decision-making algorithms provide the technology to make decisions even on partially occluded faces.

Motion Detection

Motion Detection technique provides a wide range of application fields - from the new generation of game control interfaces to modern-day security and monitoring systems. SeeStorm motion detection technique summarizes all known motion detection technologies to detect motion clusters, estimate their characteristics and generate excitation signal.