Virtual Space Gaze and Emotion Inference With Lower Calculation Load

The patent extracts only the necessary information for line-of-sight estimation from the captured image, rather than performing comprehensive image processing. By focusing on specific features (such as eye position and orientation) and using simplified geometric calculations based on camera parameters, the system achieves adequate line-of-sight estimation without the heavy computational burden of full image analysis.

The patent performs preliminary setup by storing camera parameters (focal length, optical center, etc.) before line-of-sight detection is needed. During actual detection, the system uses these pre-prepared parameters in simple geometric calculations rather than performing complex image processing, significantly reducing processing time while maintaining detection accuracy.

The patent segments the virtual space into multiple regions or objects of interest, and performs emotion estimation specifically toward these segmented targets. By dividing the complex task of monitoring the entire virtual environment into smaller, focused estimations toward individual objects, the system achieves accurate emotion detection without overwhelming system complexity.