Title :
Low bit rate coding of image sequences using neural networks
Author :
Kalogeras, Dimitrios ; Kollias, Stefanos
Author_Institution :
Div. of Comput. Sci., Nat. Tech. Univ. of Athens, Greece
Abstract :
Neural networks are proposed in this paper as an efficient means for extending international image image sequence coding standards to achieve low bit rates for transmission in modern communication systems. In particular a structured neural network architecture is proposed for adaptively selecting regions of interest (ROI) in the images to be coded; high compression is obtained by using different quantization in each selected region region category. Simulation results illustrate the performance of the technique using videoconference sequences
Keywords :
image coding; image sequences; learning (artificial intelligence); neural net architecture; neural nets; international image image sequence coding standards; low bit rate coding; modern communication syste; neural networks; structured neural network architecture; videoconference sequences; Bit rate; Head; Image coding; Image reconstruction; Image sequences; Layout; Neural networks; Quantization; Video compression; Videoconference;
Conference_Titel :
Neural Networks, 1995. Proceedings., IEEE International Conference on
Conference_Location :
Perth, WA
Print_ISBN :
0-7803-2768-3
DOI :
10.1109/ICNN.1995.488877
