COMP9519: Multimedia Systems

Assignment 1 (Project)


1.     Latest Specification

       The latest specification of Assignment 1 is here .  

       The Stefan sequence in YUV 4:2:0 is located here.

       Due date: 4:30 PM 7 Sept 2009

 

1.1 Change Log

  10 Aug 2009:   FAQ updated.

  6 Aug 2009:   Assignment page setup.


2.     Skeleton File

     Here are the skeleton files for you to start with.   

The template C code gives you the format of command line arguments and the hints for reading YUV 4:2:0 data. Please read the comments to make yourself clear of the input output format before your start implementation. You can change the structure of the file, as long as it follows the same argument formats and output formats.

Also include is a MS Excel file “ass1graphs.xls” for your plots.


3.     YUV File Viewer

For Linux:  yay

You can check it out through the link here, or get a local copy .

The yay viewer is already installed in the class account, what you need to do is type

9519

under the CSE server, and then followed by

yay -s <width>x<height> filename.yuv

For Windows: YUVviewer

      Check it out.


4.     FAQ

4.1 What is the format of the generated histogram file?

After your program is invoked, four CSV (“Comma Separated Value”) files would be generated.  The commas separate the pixel value and frequency. One example of *.hist.csv file can be found here. For *.psnr.csv, one example can be found here. Note that CSV files can be directly opened in MS Excel and will help you generate the graphs.

4.2 How do you define the search range for macro blocks at the

edges of the frame?

It will be the same as other macro blocks except the area of the search range that is not within the frame bounds will be ignored.

4.3 Do we need to use the U & V components?

For this assignment, just use the Y component.

4.4 What should the PSNR calculations be based on?

The PSNR calculation should be based on the frames from uncompressed sequence and the corresponding MC frames generated from the ME algorithm. The result should be a PSNR value for each frame.