Robust and efficient visual motion detection based on Fast Fourier Transform and phase-based algorithms

This technology is an efficient and robust algorithm that can be used for visual motion detection and image reconstruction.

Unmet Need: Fast and efficient algorithm for reliable motion detection

Conventional real-time visual motion detection software solutions often struggle to reliably detect motion when lighting conditions are poor or image quality is low. Current software is also computationally expensive and relatively inefficient. Therefore, there is a need for fast and reliable motion detection technology that is compatible with modern computing devices.

The Technology: Robust and reliable method for visual motion detection using Fast Fourier Transform (FFT)

This technology is an efficient and reliable method for visual motion detection that performs well in a variety of lighting conditions and noisy environments. Instead of conventional amplitude or optical flow-based approaches, it uses a phase-based algorithm to reconstruct images and detect motion. Consequently, this technology remains reliable even in exceptionally low-contrast and high-noise conditions. It can be easily implemented in most current computing devices and applied to economic pricing movement models, geologic seismic monitoring, and neural network analysis.

This technology has been successfully validated using video surveillance footage with low contrast and high noise. It is currently being tested in other applications, such as ego-motion inference and neural circuits.

Applications:

• Video security and surveillance software
• Entertainment applications, including Virtual Reality, Augmented Reality, and Motion Capture
• Film and video editing, including visual effects software
• Machine learning and deep learning
• High-dimensional simulation applications in chemistry, geology, finance, and meteorology
• Motion tracking for cell phone camera software
• Long-range missile defense and tracking
• Automated pilot control in planes, cars, and unmanned or armed vehicles

Advantages:

• Computationally fast and efficient
• Common FFT-based algorithm for compatibility with modern computing devices
• Robust and accurate visual motion detection in extreme conditions (low contrast, high noise)
• Applicable to various industries such as financial markets, geology, neurology, etc.

Lead Inventor:

Aurel A. Lazar, Ph.D.

Related Publications:

• Lazar, A. A., Ukani, N. H, Zhou, Y., “A Motion Detection Algorithm Using Local Phase Information,” Computational Intelligence and Neuroscience, vol. 2016, Article ID 7915245, 2016.

Tech Ventures Reference:

• IR CU15252, CU14297

• Licensing Contact: Greg Maskel