Combining Motion Compensation with Spatiotemporal Constraint for Video Deblurring.

Area of Science:

• Computer Vision
• Image Processing
• Video Restoration

Background:

• Camera shake often results in blurry videos, degrading visual quality.
• Existing video deblurring methods struggle with motion blur and temporal inconsistencies.

Purpose of the Study:

• To develop an effective video deblurring method for camera shake.
• To leverage spatiotemporal information for improved blur kernel estimation and frame restoration.

Main Methods:

• Motion vector estimation and compensation using previous restored frames.
• Improved blur kernel estimation incorporating motion-compensated frames and regularization.
• Spatiotemporal constraint algorithm with temporal regularization for enhanced consistency and artifact suppression.
• Extended Fast Total Variation de-convolution (FTVd) for energy function minimization.

Main Results:

• Successfully restored blurry video frames caused by camera shake.
• Achieved enhanced temporal consistency and reduced noise and ringing artifacts.
• Demonstrated state-of-the-art performance in both subjective and objective evaluations.

Conclusions:

• The proposed method effectively restores videos with camera shake.
• Integration of motion compensation and spatiotemporal constraints significantly improves deblurring quality.
• The technique offers a robust solution for practical video restoration applications.