lect916h - 1 Model-Based Video Coding Model-Based...

Info iconThis preview shows pages 1–9. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 1 Model-Based Video Coding Model-Based Compression Object-based Knowledge-based Semantic-based 2 Model-Based Compression Analysis Synthesis Coding Video Compression MC/DCT Source Model: translation motion only Encoded Information: Motion vectors and color of blocks Object-Based Source Model: moving unknown objects translation only affine affine with triangular mesh Encoded Information: Shape , motion, color of each moving object 3 Video Compression Knowledge-Based Source Model: Moving known objects Encoded Information: Shape, motion and color of known objects Semantic Source Model: Facial Expressions Encoded Information: Action units Object Segmentation Frame k-1 Frame k 4 Object Segmentation Segmentation based on single frame (static) Motion-based segmentation Optical flow based Compute optical flow Cluster optical flow into regions Change detection Threshold consecutive frame difference Determine connected components Estimate motion for each connected component Determine motion failures Iterate Simultaneous motion estimation and segmentation Tian & Shah optical flow http://www.cs.ucf.edu/~vision/papers/shah/95/TIS95.pdf 5 Object-Based Coding Frame Unchanged region changed region Uncovered background Moving region Objec-1 Objec-2 Objec-3 MC MC MF MC MF MF Detection of Uncovered Background k f 1 + k f CD(k,k+1) 6 Detection of Uncovered Background All pixels in frame k+1 that are changed, are traced back to the frame k, using inverse of motion vectors. If the inverse of motion vector points to a pixel in frame k, which is within the changed region than it is a moving pixel; otherwise it is an uncovered background pixel. Frame k-1 Frame k 7 2-D objects With Affine Motion Analysis Segment image by change detection Compute motion parameters, e.g. affine (x=Ax+b) Synthesize the region in the current frame using previous frame and the motion parameters if the difference between actual and synthesized region is significant, recursively segment the region into small regions 2-D objects With Affine Motion Synthesis Using final segmented regions and the motion parameters synthesize the frame, and compute the synthesis error. Coding Code motion parameters (using 6 to 7 bits each) Code region shapes Code prediction errors 8 Affine Transformation With Triangular Meshes Partition the current frame into triangular patches....
View Full Document

This note was uploaded on 06/12/2011 for the course CAP 6411 taught by Professor Shah during the Spring '09 term at University of Central Florida.

Page1 / 28

lect916h - 1 Model-Based Video Coding Model-Based...

This preview shows document pages 1 - 9. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online