Multiple descent cost competitive learning and data-compressed 3-D morphing

Author

Matsuyama, Yasuo ; Shimazu, Takashi ; Matsuo, Go ; Arisaka, Takeshi

Author_Institution

Dept. of Electr., Electron. & Comput. Eng., Waseda Univ., Tokyo, Japan

Volume

1

fYear

1999

fDate

1999

Firstpage

375

Abstract

Multiple descent cost competitive learning is applied to data-compressed texture generation for 3D image processing and graphics. This learning method organizes itself by generating two types of feature maps: the grouping feature map and the weight vector feature map, which can both change regional shapes. This merit makes it possible for users to generate data-compressed image morphing. The resulting textures can be used to create virtual 3D objects. Examples are given of generating emotional expressions. The theoretical relationship between the α-EM (expectation maximization) algorithm and the multiple descent cost competitive learning algorithm is also clarified

Keywords

gradient methods; image coding; image morphing; image texture; self-organising feature maps; unsupervised learning; vector quantisation; vectors; α-EM algorithm; 3D graphics; 3D image processing; data-compressed 3D image morphing; data-compressed texture generation; emotional expression generation; expectation maximization; grouping feature map; multiple descent cost competitive learning; regional shape change; self-organizing learning method; virtual 3D objects; weight vector feature map; Computer graphics; Costs; Data compression; Data engineering; Image generation; Image processing; Learning systems; Shape; Terminology; Vector quantization;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Information Processing, 1999. Proceedings. ICONIP '99. 6th International Conference on

Conference_Location

Perth, WA

Print_ISBN

0-7803-5871-6

Type

conf

DOI

10.1109/ICONIP.1999.844017

Filename

844017