Lecture-12 - Lecture-12 Video Compression Model-based Video...

Info iconThis preview shows pages 1–5. Sign up to view the full content.

View Full Document Right Arrow Icon
Copyright Mubarak Shah 2003 Lecture-12 Model-based Video Compression Li, Teklap Video Compression What is Compression? ! Compression is a process of converting data into a form requiring less space to store or less time to transmit, which permits the original data to be reconstructed with acceptable precision at a later time. Orange Juice Analogy! ! Freshly squeezed orange juice (uncompressed) ! Remove water (redundancy), convert it to concentrate (encoding) ! Shipped, stored, and sold. ! Add water to concentrate (decoding), tastes like freshly squeezed!!!
Background image of page 1

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

View Full DocumentRight Arrow Icon
Why is compression necessary? ! Storage space limitations ! Transmission bandwidth limitations. Why is compression acceptable? ! Limitations of visual perception " Number of shades (colors, gray levels) we can perceive " Reduced sensitivity to noise in high- frequencies (e.g. edges of objects) " Reduced sensitivity to noise in brighter areas ! Ability of visual perception " Ability of the eye to integrate spatially " Ability of the mind to interpolate temporally Why is compression acceptable? ! Some type of visual information is less important than others ! Goal is to throw away bits in psycho-visually lossless manner ! We have been conditioned to accept imperfect reproduction ! Limitations of intended output devices Why is compression possible? ! Some sample values (gray levels, colors) are more likely to occur at a particular pixel than others. " Remove spatial and temporal redundancy that exist in natural video ! Correlation itself can be removed in a lossless fashion ! Important to medical applications ! Only realizes about 2:1 compression
Background image of page 2
Why is compression possible? ! No single algorithm can compress all possible data ! Random data cannot be compressed Lossless Compression ! Needed when loss is unacceptable or highly undesirable ! Fixed compression ratio is hard to achieve ! Compression/decompression time varies with image Lossy Compression ! Used when loss is acceptable or inevitable ! Permits fixed compression ratios ! Better suited for fixed time decompression Compression Techniques ! Subsampling ! Quantization ! Delta Coding ! Prediction ! Color space conversion ! Huffman coding ! Run-length encoding ! De-correlation ! Motion Compensation ! Model-based compression
Background image of page 3

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

View Full DocumentRight Arrow Icon
Subsampling ! Selecting one single value to represent several values in a part of the image. " For example, use top left corner of 2X2 block to represent the block " Compression ratio 75% 11 15 19 55 13 14 21 32 39 17 24 76 43 34 27 80 11 11 19 19 11 11 19 19 39 39 24 24 39 39 24 24 Subsampling ! A better way- averaging ! Compression ratio 75% 11 15 19 55 13 14 21 32 39 17 24 76 43 34 27 80 13 13 32 32 13 13 32 32 33 33 51 51 33 33 51 51 Quantization ! Mapping of a large range of possible sample values into a smaller range of values or codes. ! Fewer bits are required to encode the quantized sample. ! Examples " -Letter grades (A, B, C, D, F) " Rounding of person’s age, height, or weight Quantization !
Background image of page 4
Image of page 5
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 10/04/2011 for the course CAP 6411 taught by Professor Shah during the Spring '09 term at University of Central Florida.

Page1 / 24

Lecture-12 - Lecture-12 Video Compression Model-based Video...

This preview shows document pages 1 - 5. Sign up to view the full document.

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