Invention:
This technology is a novel method for object detection in environments with significant scattering and other forms of interference. The approach leverages maximum likelihood estimation principles to achieve optimal object detection and characterization based on known physical models of scattering and emission from thermal emission, light scattering, or other detectable signatures. For any given signal, the likelihood that it was generated by interference or background noise is compared to the likelihood that it was generated by the presence of an object. These likelihood calculations are performed by analyzing signals against physical models of scattering and emission phenomena. This detection method can estimate the locations, shapes, and velocities of unknown objects and is particularly relevant in drone detection, medical imaging, and other fields where detection is prone to scattering and interference. This method offers reliable, accurate results while remaining computationally efficient.
Background:
Current object detection and characterization methods generally center around the use of neural networks. However, these neural network-based approaches often aren’t adaptable to a wide range of physical situations, sometimes have local extrema issues, and come with computational inefficiencies. This alternative method provides highly efficient and reliable object detection and characterization.
Applications:
- Aerospace
- Defense
- Healthcare / Medical Imaging
- Optics
- Environmental monitoring
Advantages:
- Physics-based solution that uses smaller training data sets
- Increased computational efficiency
- Enhanced reliability in detection
- Versatile; can be used in many fields
