Video Stabilization: Camera Stabilization Based on 2.5D Motion Estimation and Inertial Motion Filtering

Project Status: Complete (Years: 1995 – 2000)

Research Description

We have presented a novel approach to stabilize digital video sequences. While most of the real-time image stabilization systems used 2D motion models, a 2.5D inter-frame motion model was proposed so that the stabilization system can work in situations where significant depth changes are present and the camera has both rotational and translational movements. By studying the physical mechanism of a vehicle’s motion, we designed an inertial motion filtering model in order to eliminate the vibration of the video sequences and to achieve good perceptual properties. The implementation of this new approach integrated four modules: pyramid-based motion detection, motion classification and 2.5D motion parameter estimation, inertial motion filtering, and affine-based motion compensation. Among them the concept of motion classification was the first attempt among researchers working in image stabilization to adapt the algorithm to different kinds of motion during a mission. Our stabilization system can smooth unwanted vibrations of video sequences in real-time. The implemented system on IBM PC compatible machines have shown that his algorithm outperforms many algorithms that require parallel pipeline image processing machines.

Research Description

Sponsors