9/29 Notes Multimedia Representation and Storage

"Data Compression"
Also known as source coding.
Encoding information using fewer bits (or other units) than unencoded representation.
Source coding only works if both the sender and receiver understand the code.

Helps reduce cost assoc. with hard disk space and transmission bandwith.
Probs: compressed data needs to be decompressed to be used, need storage space, expensive equipment, distortion.

Lossless versus lossy compression
Exploits statistical redundancy. Lossless compression algorhythyms fan fail to suppress some files if a pattern is not recognized.

Lossy-aka perceptual encoding.
Lossless 25.888888888
Compressed: 29[9]8
Lossy: 26

Application-optimizes disk space in office computers.
In lossy audio compression, psycoacoustics are used to remove nonaudible (less audible) components of a signal.

Theory
Lossless data compression systems can be described as a four-step model; lossy has even more, including prediction,frequency, transformation, and quantization.
Lempel-Ziv (LZ) compression methods are the most popular algorhythms for lossless storage.
DEFLATE-variation on LZ which is optimized for decompression speed and compression ratio-compression can be slow.
LZW-LZ Welch is used for GIF.
LZRenau-for Xip methods.
Best compressors-probalisitc models.

"Data Compression Basics"-covers run-length encoding and gives more examples of compression.

"Imagining Pittsburgh"
The article focuses on a massive project undertaken by Pitts DRL to combine over 20 photo collections from Carnegie Museum of Art and the Historical Society of Western Pennsylvania. The article covers communication and selection challenges which encompass a project undertaken by several different organizations with different goals of what the project should accomplish. A document was created to determine what should go into the collection based characteristics such as size, format and condition. Standards were set to establish minimum image quality. Generic copyright ws chosen. To prevent overlap of images, subject headings were used. Split collections still remain a problem and plans are being made for collaboration between organizations.

Metadata challenges were met by using Dublin Core characteristics and choosing a catalog to group the images.

"YouTube and Libraries"
The article focuses on utilizing YouTube for distance learning, providing access to classes.