Traditional video capture devices use analog-to-digital conversion and represent information in the amplitude domain. But the demand for smaller, low-voltage circuitry is making it increasingly difficult to implement these methods. Time encoding machines (TEMs) offer a promising alternative by representing information in the time domain at irregularly spaced times. This asynchronous approach transforms analog signals to discrete codes, enabling low-power video capture. Time decoding machines (TDMs) reconstruct these encoded visual signals. These TEMs and TDMs can be utilized to capture 3D and stereoscopic color video or imagery. This technology has a wide range of applications, from providing color vision for visually-impaired individuals to enhancing computer vision for robotics.
This technology implements neural circuitry, a scheme representative of neurons in the brain, to sample and encode color visual signals. Using TEMs and TDMs, analog signals are encoded as an asynchronous time sequence and then reconstructed in digital form. This method allows for frameless encoding of video, which can theoretically allow video to be captured and replayed at any desirable frame rate. Furthermore, the neural circuitry underlying the technology facilitates robust, parallel encoding. This circuitry can also be configured randomly, which allows for ease of design and manufacturing, especially for low-power devices.
A demonstration of this technology has shown that analog video streams can successfully be encoded and recovered at near perfection without any noticeable error.
Patent Pending (US 20140267606)
Tech Ventures Reference: IR CU13246