Neural network based motion vector computation and application to MPEG coding

Author

Skrzypkowiak, S.S. ; Jain, V.K.

Author_Institution

Dept. of Electr. Eng., Univ. of South Florida, Tampa, FL, USA

Volume

2

fYear

1994

fDate

13-16 Nov 1994

Firstpage

938

Abstract

A neural network based motion-estimation technique is developed, that is applicable to sub-pixel as well as large movements. Experiments on typical test frame sequences indicate marked improvement in accuracy of motion vector estimates over the MPEG logarithmic block matching algorithm. The method utilizes a modified Hopfield neural network. Due to the neural network´s fault-tolerant nature and parallel computation capability, fast, accurate, and reliable results are obtained. Application to MPEG based video compression is also discussed

Keywords

Hopfield neural nets; data compression; image sequences; motion estimation; parallel algorithms; video coding; MPEG based video compression; MPEG coding; fault-tolerant; large movements; modified Hopfield neural network; motion-estimation technique; neural network based motion vector computation; parallel computation; sub-pixel movements; test frame sequences; Computer applications; Computer networks; Concurrent computing; Fault tolerance; Hopfield neural networks; Motion estimation; Neural networks; Testing; Transform coding; Video compression;

fLanguage

English

Publisher

ieee

Conference_Titel

Image Processing, 1994. Proceedings. ICIP-94., IEEE International Conference

Conference_Location

Austin, TX

Print_ISBN

0-8186-6952-7

Type

conf

DOI

10.1109/ICIP.1994.413493

Filename

413493