UCL - Communications and Remote Sensing Laboratory (TELE)

M2VTS logo

M2VTS Multimodal Face Database
Release 1.00

ACTS Project 102

Image files

Originally, the images were recorded using a CCIR 601 format :

In order to lower the memory requirement, the images are down-converted into the CIF format by removing one field out of two and horizontally down-sampling the remaining fields. This results in a : The down-sampling low-pass filter being used is listed below : 
     -0.1133   0   0.3437   0.539   0.3437   0   -0.1133
At last, by keeping the active lines only, the final format for the database is : Sequences of images are stored in .lcc files, using the following structure : 
   begin of file

         Image 1 - luminance (286x350 bytes)
         Image 1 - Cb (286x175 bytes)
         Image 1 - Cr (286x175 bytes)
         Image 2 - luminance (286x350 bytes)
         Image 2 - Cb (286x175 bytes)
         Image 2 - Cr (286x175 bytes)
         Image 3 - ...
         ...

   end of file
Most signal processing can be done using the Lum/Cb/Cr components. However, this format may be not recognized by your favourite image viewer. You will find here a little c++ program that can be used to convert an image from a .lcc sequence into SUN rasterfile .ras. Its usage is the following :

makeRAS sequence.lcc frame#
As output, a temp.ras file is created. This file can be viewed using the XV application for example. This code also shows how to handle the Lum/Cb/Cr 4:2:2 format and convert it into R/G/B components.

If you need some more advise about color space conversion, take a look at Charles Poynton' site : Poynton's Colour FAQ

Go to next section...
Back to main page...

Last modified Decembre 18, 1996.
Author: Stéphane Pigeon