A Comprehensive Motion Estimation Technique for the Improvement of EIS Methods Based on the SURF Algorithm and Kalman
Xuemin Cheng1, Qun Hao2, Mengdi Xie3
1Graduate School at Shenzhen, Tsinghua University, Shenzhen 518055, China. cheng-xm@mail.tsinghua.edu.cn.
Sensors (Basel, Switzerland)
|April 13, 2016
Summary
This study introduces a new video stabilization method using Speeded Up Robust Features (SURF), RANSAC, and Kalman filters. The technique effectively stabilizes videos with large vibrations, improving image quality.
Area of Science:
- Computer Vision
- Image Processing
- Signal Processing
Background:
- Video stabilization is crucial for removing unwanted motion in digital videos.
- Existing methods often struggle with large vibrations and accumulated errors.
Purpose of the Study:
- To develop a comprehensive motion estimation method for electronic image stabilization.
- To enhance video stabilization by integrating SURF, RANSAC, and Kalman filters, considering scaling, translation, and rotation.
Main Methods:
- Utilized Speeded Up Robust Features (SURF) for sub-pixel feature point detection and matching.
- Employed modified Random Sample Consensus (RANSAC) to eliminate false matches.
- Integrated a Kalman filter model with modified cascading parameters for motion and scaling estimation, and adjacent frame compensation.
Main Results:
- Achieved significant reduction in accumulated errors across frames.
- Improved Peak Signal to Noise Ratio (PSNR) by 8.2 dB.
- Demonstrated effective stabilization even with large video vibration amplitudes.
Conclusions:
- The proposed method offers robust video stabilization by accurately estimating global motion and compensating for camera movement and scaling.
- The integration of SURF, RANSAC, and Kalman filtering provides a powerful solution for electronic image stabilization.
