THERMAL RESOLUTION ENHANCEMENTS USING ARTIFICIAL INTELLIGENCE AND SPECTRAL EMISSIONS

Smart overview of the Invention

The patent application outlines a method for enhancing thermal imaging resolution through the use of artificial intelligence and spectral emissions. It involves obtaining a readout from a thermal imaging sensor that captures long-wave infrared (LWIR) light across different wavelength ranges. Multiple images are generated from this readout, each reflecting the same scene but using different wavelength ranges. These images are used to identify objects and their LWIR profiles, which are then matched with saved profiles to determine the object's type. The method also includes colorizing thermal images based on the identified object types.

Background

Extended Reality (ER) systems, including virtual, augmented, and mixed reality platforms, are increasingly popular for providing immersive experiences. These systems can display both virtual and real-world environments using head-mounted devices (HMDs) or other computing systems. ER systems can present holograms, which are virtual images that appear as either 2D or 3D objects. The technology aims to enhance the realism of these holograms, making them appear as part of the real world. While the background section provides context, the disclosed invention is not limited to these specific environments.

Technical Details

The invention employs a thermal imaging sensor with multiple pixel sets, each capturing different LWIR wavelength ranges. Images from these sensors are analyzed to identify objects and their thermal profiles. By matching these profiles with stored data, the system can determine object types and apply colorization to thermal images accordingly. The technology leverages machine learning models to extract subtle spectral differences, improving object detection and classification. Additionally, it can be implemented using advanced optical elements like diffractive or meta-surface lenslets.

Challenges Addressed

Integrating thermal cameras into ER systems presents challenges such as bulkiness and battery consumption, which can affect user comfort. Traditional thermal cameras also struggle with resolution issues due to spectral signature blending. Previous machine learning approaches have been insufficient in resolving these issues. The proposed method addresses these challenges by enhancing thermal image resolution and enabling better object distinction through AI-driven spectral analysis.

Potential Applications

The invention's ability to improve thermal imaging resolution and object classification has broad applications in ER systems. It enhances the user's ability to perceive both visible and thermal content, offering a richer and more immersive experience. The technology can be particularly beneficial in fields requiring precise thermal imaging, such as security, surveillance, and various industrial applications. Overall, it represents a significant advancement in integrating thermal imaging with extended reality technologies.