logging in or signing up Compression Techniques ankush85 Download Post to : URL : Related Presentations : Share Add to Flag Embed Email Send to Blogs and Networks Add to Channel Uploaded from authorPOINT lite Insert YouTube videos in PowerPont slides with aS Desktop Copy embed code: Embed: Flash iPad Dynamic Copy Does not support media & animations Automatically changes to Flash or non-Flash embed WordPress Embed Customize Embed URL: Copy Thumbnail: Copy The presentation is successfully added In Your Favorites. Views: 10275 Category: Education License: All Rights Reserved Like it (13) Dislike it (1) Added: March 05, 2009 This Presentation is Public Favorites: 5 Presentation Description No description available. Comments Posting comment... By: ohol.chetan (34 month(s) ago) send this ppt to firstname.lastname@example.org these ppts are really usefull to me Saving..... Post Reply Close Saving..... Edit Comment Close By: shwetavarshney05 (35 month(s) ago) i like it Saving..... Post Reply Close Saving..... Edit Comment Close Premium member Presentation Transcript Media Compression Techniques : Media Compression Techniques Table of Contents : Table of Contents Image Compression Methods JPEG GIF 89a Wavelet Compression Fractal Sound Compression MPEG Audio Overview MPEG Layer-3 (MP3) MPEG AAC Video Compression Methods H.261 MPEG/MPEG-2 MPEG-4 MPEG-7 JPEG Compression: Basics : JPEG Compression: Basics Human vision is insensitive to high spatial frequencies JPEG Takes advantage of this by compressing high frequencies more coarsely and storing image as frequency data JPEG is a “lossy” compression scheme. Losslessly compressed image, ~150KB JPEG compressed, ~14KB Digital Image Representation : Digital Image Representation JPEG can handle arbitrary color spaces (RGB, CMYK, YCbCr (separates colors into grayscale components) Luminance/Chrominance commonly used, with Chrominance subsampled due to human vision insensitivity Uncompressed spatial color data components are stored in quantized values (8, 16, 24bit, etc). Flow Chart of JPEG Compression Process : Flow Chart of JPEG Compression Process Divide image into 8x8 pixel blocks Apply 2D Fourier Discrete Cosine Transform (FDCT) Transform Apply coarse quantization to high spatial frequency components Compress resulting data losslessly and store 8x8 pixel blocks FDCT Frequency Dependent quantization Zig-zag scan Huffman encoding JPEG syntax generator Quantization Table output Example of Frequency Quantization with 8x8 blocks : Example of Frequency Quantization with 8x8 blocks Quantization Matrix to divide by Color space values (spatial data) Quantized spatial frequency values Color space values (spatial data) Scanning and Huffman Encoding : Scanning and Huffman Encoding Spatial Frequencies scanned in zig-zag pattern (note high frequencies mostly zero) Huffman encoding used to losslessly record values in table 0,2,1,-1,0,0,1,0,1,1,0,0,1,0,0,0,-1,0,0,… 0 Can be stored as: (1,2),(0,1),(0,-1),(2,1),(1,1),(0,1),(0,1),(2,1),(3,1),EOB Examples of varying JPEG compression ratios : Examples of varying JPEG compression ratios 500KB image, minimum compression 40KB image, half compression 11KB image, max compression Close-up details of different JPEG compression ratios : Close-up details of different JPEG compression ratios Uncompressed image (roughness between pixels still visible) Half compression, blurring & halos around sharp edges Max compression, 8-pixel blocks apparent, large distortion in high-frequency areas JPEG Encoding modes : JPEG Encoding modes Sequential mode Image scanned in a raster scan with single pass, 8-bit resolution Sequential mode Step-by-step buildup of image from low to high frequency, useful for applications with long loading times (internet, portable devices, etc) Hierarchical mode Encoded using low spatial resolution image and encoding higher resolution images based on interpolated difference, for display on varying equipment GIF 89a Image Compression : GIF 89a Image Compression Compuserve’s image compression format Best for images with sharp edges, low bits per channel, computer graphics where JPEG spatial averaging is inadequate Usually used with 8-bit images, whereas JPEG is better for 16-bit images. GIF 89a examples vs. JPEG : GIF 89a examples vs. JPEG GIF Image, 7.5KB, optimal encoding JPEG, blotchy spots in single-color areas Wavelet Image Compression : Wavelet Image Compression Optimal for images containing sharp edges, or continuous curves/lines (fingerprints) Compared with DCT, uses more optimal set of functions to represent sharp edges than cosines. Wavelets are finite in extent as opposed to sinusoidal functions Several different families of wavelets. Source: “An Introduction to Wavelets”. http://www.amara.com/IEEEwave/IEEEwavelet.html#contents Wavelet vs. JPEG compression : Wavelet vs. JPEG compression Wavelet compressionfile size: 1861 bytescompression ratio - 105.6 Source: “About Wavelet Compression”. http://www.barrt.ru/parshukov/about.htm. JPEG compression file size: 1895 bytescompression ratio - 103.8 Wavelet compression advantages : Wavelet compression advantages Fig. 1. Fourier basis functions, time-frequency tiles, and coverage of the time-frequency plane. Fig. 2. Daubechies wavelet basis functions, time-frequency tiles, and coverage of the time-frequency plane Source: “An Introduction to Wavelets”. http://www.amara.com/IEEEwave/IEEEwavelet.html#contents Fractal Based Image Compression : Fractal Based Image Compression Image compressed in terms of self-similarity rather than pixel resolution Can be digitally scaled to any resolution when decoded Table of Contents : Table of Contents Image Compression Methods JPEG GIF 89a Wavelet Compression Fractal Sound Compression MPEG Audio Overview MPEG Layer-3 (MP3) MPEG AAC Video Compression Methods H.261 MPEG/MPEG-2 MPEG-4 MPEG-7 MPEG Audio basics & Psychoacoustic Model : MPEG Audio basics & Psychoacoustic Model Human hearing limited to values lower than ~20kHz in most cases Human hearing is insensitive to quiet frequency components to sound accompanying other stronger frequency components Stereo audio streams contain largely redundant information MPEG audio compression takes advantage of these facts to reduce extent and detail of mostly inaudible frequency ranges MPEG-Layer3 Overview : MPEG-Layer3 Overview MP3 Compression Flow Chart MPEG Layer-3 performance : MPEG Layer-3 performance MPEG-2 Advanced Audio Coding (AAC) codec (next generation) : MPEG-2 Advanced Audio Coding (AAC) codec (next generation) Sampling frequencies from 8kHz to 96kHz 1 to 48 channels per stream Temporal Noise Shaping (TNS) smooths quantization noise by making frequency domain predictions Prediction: Allows predictable sound patterns such as speech to be predicted and compressed with better quality MPEG-2 AAC Flowchart : MPEG-2 AAC Flowchart Table of Contents : Table of Contents Image Compression Methods JPEG GIF 89a Wavelet Compression Fractal Sound Compression MPEG Audio Overview MPEG Layer-3 (MP3) MPEG AAC Video Compression Methods H.261 MPEG/MPEG-2 MPEG-4 MPEG-7 Video Compression with Temporal Redundancy : Video Compression with Temporal Redundancy Using strictly spatial redundancy (JPEG) gives video compression ratios from 7:1 to 27:1 Taking advantage of temporal redundancy in video gives 20:1 to 300:1 compression for H.261, or 30:1 to 100:1 for high quality MPEG-2 Videoconferencing Compression with H.261 : Videoconferencing Compression with H.261 H.261 is standard recommended for videoconferencing over ISDN lines. Takes advantage of both spatial and temporal redundancy in moving images Extremely similar to JPEG, but uses initial frame plus motion vectors to predict subsequent frames H.261 Block Structure : H.261 Block Structure Basic unit of processing is in 8x8 pixel blocks. Macro Blocks (MB, 16x16 pixels) are used for motion estimation, 4 blocks of luminance, 2 of chrominance Groups of Blocks (GOB) of 3x11 MB’s are stored together with a header in stream. H.261 Block Structure of bitstream : H.261 Block Structure of bitstream Source: “H.261 Videoconferencing Codec” http://www.uh.edu/~hebert/ece6354/H261-report.pdf Block structure of H.261 video bitstream, Common Intermediate Format (CIF), 360x288 pixels luminance, 180x144 pixels chrominance H.261 Decoding (Similar to encoding process) : H.261 Decoding (Similar to encoding process) Encoded Bitstream Bitstream DEcoder Loop Filter Inverse Quantizer IDCT Decompressed Video Motion Compensation Reference Frame MPEG Video Compression : MPEG Video Compression Supports JPEG and H.261 through downward compatibility Supports higher Chrominance resolution and pixel resolution (720x480 is standard used for TV signals) Supports interlaced and noninterlaced modes Uses Bidirectional prediction in “Group Of Pictures” to encode difference frames. Source: “Parallelization of Software Mpeg Compression” http://www.evl.uic.edu/fwang/mpeg.html “Group Of Pictures” inter-frame dependencies in a stream MPEG 1 & 2 Bitstream : MPEG 1 & 2 Bitstream Source: http://www.doc.ic.ac.uk/~nd/surprise_96/journal/vol4/sab/report.html The MPEG data hierarchy MPEG-4 : MPEG-4 Original goal was for 10 times better compression than H.261 Goals shifted to Flexible bitstreams for varying receiver capabilities Stream can contain new applications and algorithms Content-based interactivity with data stream Network independence (used for Internet, Wireless, POTS, etc) Object based representations MPEG-4 audio-visual scene composition : MPEG-4 audio-visual scene composition Can place media objects anywhere in a scene Apply transforms to change appearance or qualities of an object Group objects to form compound objects Apply streamed data to objects Interactively change viewer’s position in the virtual scene http://www.iis.fraunhofer.de/amm/techinf/mpeg4/mp4_overv.pdf MPEG-4 “Audiovisual Scene” Example : MPEG-4 “Audiovisual Scene” Example Source: “MPEG-4 Overview” http://www.chiariglione.org/mpeg/standards/mpeg-4/mpeg-4.htm MPEG-7 : MPEG-7 Media tagging format for doing searches on arbitrary media formats via feature extraction algorithms Visual descriptors such as: Basic Structures Color Texture Shape Localization of spatio-temporal objects Motion Face Recognition Audio descriptors such as : Sound effects description Musical Instrument Timbre Description Spoken Content Description Melodic Descriptors (search by tune) Uniform Silence Segment Example application: Play a few notes on a keyboard and have matched song retrieved. Conclusion : Conclusion Media compression is indispensable even as storage and streaming capacities increase Future goals oriented towards increasing ease of access to media information (similar to google for text based information) You do not have the permission to view this presentation. In order to view it, please contact the author of the presentation.