3D SOM Leaming And Neighborhood Algorithm
- Resource Type
- Conference
- Authors
- Zhang, Xueyan; Li, Hongsong; Cheng, Fulin; Wang, Yanhua; Ai, Xinyu
- Source
- 2018 5th International Conference on Systems and Informatics (ICSAI) Systems and Informatics (ICSAI), 2018 5th International Conference on. :911-915 Nov, 2018
- Subject
- Bioengineering
Communication, Networking and Broadcast Technologies
Components, Circuits, Devices and Systems
Computing and Processing
Fields, Waves and Electromagnetics
Power, Energy and Industry Applications
Robotics and Control Systems
Signal Processing and Analysis
Three-dimensional displays
Self-organizing feature maps
Neurons
Shape
Image coding
Training
Pattern recognition
self-organizing maps
three-dimensional image coding
pattern recognition
learning algorithm
neighborhood algorithm
- Language
Learning and neighborhood algorithm are an important part of 3D SOM algorithm. The five kinds of learning algorithm and Three kinds of neighborhood shape and three kinds of neighborhood decay functions for three-dimensional self-organizing feature maps (3DSOM) algorithm were proposed in this paper. And the algorithm were applied in three-dimensional image compression coding. Experimental results show that the quadratic function learning algorithm achieved the best peak signal to noise ratio (PSNR) and the 3D orthogonal cross neighborhood shape and memory function algorithm has better peak signal to noise ratio (PSNR) and subject quality.