haojian / VISTRA

Software which demonstrates principles such as spatial and temporal redundancy, colour channel subsampling, JPEG image compression, DCT transform coding, simple motion compensated video compression and motion estimation block matching

MATLAB Image & Video Compression Demos

Demonstration applications and related code for teaching the principles and implementations of image and video compression techniques.

Please refer to the accompanying user manual for more information.

Quick Start

Begin the home demo screen with the command start in the root directory of the project. To navigate between demos use the Home/Back and Forward arrows in the screen toolbar. To show or hide more detailed elements of the demos screens use the Advanced Mode button (the green plus button).

For more detailed information on each process see the documentation in each package. Where appropriate each method should also contain references to locations on the web or in standards documents where more information can be found.

Included MATLAB Packages

The following packages are included:

• GUIs: Contains the classes that implement the user interface screens. The base class contains any shared functionality and all screens should derive from this (and chain the parent constructor).
• Subsampling: Contains methods that implement functionality related to chroma subsampling and reconstruction.
• TransformCoding: Contains methods performing stages of the DCT process.
• EntropyCoding: Contains methods for the entropy encoding/decoding of JPEG (Huffman coding).
• JPEG: Contains the JPEG encoder and decoder bodies.
• Video: A motion compensated video coder, using JPEG for intraframe coding.
• MotionEstimation: The motion estimation functionality, such as difference calculation, block matching (full and diamond search).
• Utilities: Contains helper methods, such as methods to convert logical arrays of bits into numerical values and vice-versa.
• ThirdParty: 3rd party code, such as a faster implementation of the DCT.
• UnitTests: Unit tests for all packages.

License & Authors

Copyright 2014, The University Of Bristol See License.m for the project code license.

Authors:

• Stephen Ierodiaconou, Aaron Zhang, Alex Mackin and Dave Bull Third party components are copyright their respective owners:

• N-D discrete cosine transform - Andriy Myronenko, 2010

Bristol University Visual Information Laboratory

The Visual Information Laboratory exists to undertake innovative, collaborative and interdisciplinary research resulting in world leading technology in the areas of computer vision, image and video communications, content analysis and distributed sensor systems.