Temporal Video Filtering and Exposure Control for Perceptual Motion Blur

Area of Science:

• Computer Vision
• Human-Computer Interaction
• Digital Image Processing

Background:

• Traditional video cameras capture motion blur based on physical shutter mechanisms.
• Existing methods for simulating motion blur often lack perceptual realism.
• Understanding human visual perception is key to realistic digital media.

Purpose of the Study:

• To develop a novel computational method for perceptual motion blur in videos.
• To integrate predicted human eye motion into the motion blur computation process.
• To enhance the realism of motion blur for video content and artistic applications.

Main Methods:

• A post-processing technique was developed to compute perceptual motion blur.
• The method incorporates predicted eye-tracking data to inform blur computation.
• The approach was validated through a user study involving eye-tracking.

Main Results:

• The proposed technique generates perceptual motion blur closer to perceived reality compared to traditional methods.
• The system effectively simulates various shutter effects and supports artistic rendering.
• The method is computationally efficient with low additional cost for rendered content.

Conclusions:

• Incorporating eye motion significantly improves the perceptual realism of video motion blur.
• This approach offers a versatile tool for realistic video post-processing and creative applications.
• The method is practical for both real and artificial video inputs.