Slide 1Slide 2Slide 3Slide 423-1Compression23-2Application Specific Compression•Compression in an image (depending on the quality the user requests):–Count the total number of unique colors in the image and try to reduce that number to a manageable power of 2•Compression based on approximation–Assuming we’re using an RLE type scheme, runs of the same color would be beneficial–If your neighboring pixels are really close in color, maybe we should just make them the same color.–Neighbors, Regions, how do you make it look natural?–JPG is a lossy compression scheme using combinations of the above23-3Application Specific Compression•Video–Since we need a rapid succession of images (say 30 frames per second) we really don't want to store 30 bitmaps per every second of video–Even switching to 30 JPEGs (compressed images) per second doesn't help too much–We can employ further lossy and lossless techniques to video–lossy•Interlacing: store only alternating rows of each video cutting the overall size of the video in half [example on the board]–lossy or lossless•difference updates: only save what is different between successive frames not the entire frame [example on the board]23-4Application Specific Compression•Audio–Compressing audio can use any number of techniques depending on your approach to visualizing the “problem”•digital audio is a collection of discreet values over time•this attempts to create a wave of sound...–MP3 models information about the way the human ear works to throw away information that is unnecessary•E.g., can’t hear certain tones simultaneously •Also includes very clever schemes for representing waves as a series of functions instead of many individual points–MP3 saves a factor of 10, but is of course lossy. .
View Full Document
Unlocking...