Lecture_10b-09Ross

Lecture_10b-09Ross - ELEC300U: A System View of...

Info iconThis preview shows pages 1–6. Sign up to view the full content.

View Full Document Right Arrow Icon
ELEC300U: A System View of Communications: from Signals to Packets Lecture 10b Perceptual coding Lossless vs. lossy compression Perceptual models Selecting info to eliminate Quantization and entropy encoding ELEC300U 1 Some content taken with permission from material developed for the course EECS6.02 by C. Sodini, M. Perrot and H. Balakrishnan
Background image of page 1

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

View Full DocumentRight Arrow Icon
Lossless vs. Lossy Compression Huffman and LZW encodings are lossless , i.e., we can reconstruct the original bit stream exactly: bits OUT = bits IN . What we want for “naturally digital” bit streams (documents, messages, datasets, …) Any use for lossy encodings: bits OUT bits IN ? “Essential” information preserved Appropriate for sampled bit streams (audio, video) intended for human consumption via imperfect sensors (ears, eyes). Source Encoding Source Decoding Store/Retrieve Transmit/Receive bits IN bits OUT ELEC300U 2 Some content taken with permission from material developed for the course EECS6.02 by C. Sodini, M. Perrot and H. Balakrishnan
Background image of page 2
Perceptual Coding Start by evaluating input response of bitstream consumer (eg, human ears or eyes), i.e., how consumer will perceive the input. Frequency range, amplitude sensitivity, color response, … Masking effects Identify information that can be removed from bit stream without perceived effect, e.g., Sounds outside frequency range, or masked sounds Visual detail below resolution limit (color, spatial detail) Info beyond maximum allowed output bit rate Encode remaining information efficiently Use DCT-based transformations Quantize DCT coefficients Entropy code (eg, Huffman encoding) results ELEC300U 3 Some content taken with permission from material developed for the course EECS6.02 by C. Sodini, M. Perrot and H. Balakrishnan
Background image of page 3

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

View Full DocumentRight Arrow Icon
Perceptual Coding Example: Images Characteristics of our visual system opportunities to remove information from the bit stream More sensitive to changes in luminance than color spend more bits on luminance than color (encode separately) More sensitive to large changes in intensity (edges) than small changes quantize intensity values Less sensitive to changes in intensity at higher spatial frequencies use larger quanta at higher spatial frequencies So to perceptually encode image, we would need: Intensity at different spatial frequencies Luminance (grey scale intensity) separate from color intensity ELEC300U 4 Some content taken with permission from material developed for the course EECS6.02 by C. Sodini, M. Perrot and H. Balakrishnan
Background image of page 4
JPEG Image Compression JPEG = Joint Photographic Experts Group RGB to YCbCr Conversion Group into 8x8 blocks of pixels Discrete Cosine Transform Quantizer Entropy Encoder 011010… Performed for each 8x8 block of pixels ELEC300U
Background image of page 5

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

View Full DocumentRight Arrow Icon
Image of page 6
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 01/28/2011 for the course ELEC 300U taught by Professor Rossmurchandaminebermak during the Fall '08 term at HKUST.

Page1 / 22

Lecture_10b-09Ross - ELEC300U: A System View of...

This preview shows document pages 1 - 6. Sign up to view the full document.

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