Motion Compensation
Motion compensation is a set of techniques that take advantage
of redundancy in consecutive video frames. These techniques are used
in video processing applications such as video compression and video
stabilization. For both of these applications, motion compensation
is a two-step process of detection and compensation. The detection
step results in the specification of a motion vector that relates
two consecutive video frames. For video compression, the compensation
step involves using the motion vector to predict the current video
frame from the previous frame and encoding the prediction residual.
For video stabilization, the compensation step involves translating
the current frame in the opposite direction of the motion vector to
stabilize the video sequence.
The Video and Image Processing Blockset software contains a
video compression demo model that you can open by typing vipcodec at
the MATLAB command prompt.
This demo model detects motion by analyzing how much objects
move between consecutive video frames. The model aligns two sequential
video frames, subtracts them, and codes the residual.
The Video and Image Processing Blockset software also contains
a video stabilization demo model that you can open by typing vipstabilize at
the MATLAB command prompt.
The demo illustrates a motion stabilization technique based
on the sum of absolute differences (SAD) method. It applies the SAD
technique to remove unwanted translational camera motions and generate
a stabilized video.
 | Pattern Matching | | Image Compression |  |
Learn more about Simulink through this collection of videos, articles, technical literature and the Getting Started with Simulink Guide.
Get this Simulink Kit
Store 
