Abstract: Based on biological prototype of human brain and improved understanding of the functionality of the neurons and the pattern of their interconnections in the brain, many mathematical models of neural networks have been proposed to perform useful computational functions in different applications. Among these NN models, a theoretical model used to explain the information-processing characteristics of the cerebellum was developed independently by Marr (1969) and Albus (1971). CMAC, first proposed by Albus, signifies cerebellar model articulation controller. CMAC is a learning structure that imitates the organization and functionality of the cerebellum of the human brain. That model revealed the structure and functionality of the various cells and fibers in the cerebellum.

The core of CMAC is an associative memory which has the ability to realize complex nonlinear functions. CMAC takes advantage of the input-redundancy by using distributed storage and can learn nonlinear functions extremely quickly due to the local adjustment of its weights. CMAC has been applied in the closed-loop control of complex dynamical systems for the reason of good generalization capability and fast learning property.

PLENARY LECTURE 2

Boolean Functions and Contemporary Cryptography

Professor Oleksander Markovskyy
National Technical University of Ukraine
(Polytechnic Institute of Kiev)
Department of Computer Engineering

Abstract: The one way (non invertible) transformations constitute the basis of the contemporary cryptography. A substantial part of the contemporary cryptographic algorithms is based on the one way transformations, which use the Boolean functions. In particular, the cryptographic properties of two most important classes of the algorithms (block cipher and stream ciphers for symmetrical coding) are based on the Boolean transformations.
In this Plenary Lecture, the contemporary state of the Boolean functions theory will be presented in details including specific cryptographic tasks for implementation. The basic properties of the Boolean functions, which are important for the cryptographic applications, and also the contemporary state of the problem for obtaining the Boolean functions, which possess such properties, will be mentioned, analyzed and discussed. Considerable attention will be given to the theoretical issues of the development of the Boolean functions for developing and implementing effective cryptographic systems.

SESSION: Discrete Wavelet Transforms

Chair: Tze-Yun Sung, Jianyu Lin

Memory-Efficient and High-Performance Parallel-Pipelined Architectures for 5/3 Forward and Inverse Discrete Wavelet Transform	Tze-Yun Sung	560-129
Memory-Efficient and High-Perfofrmance 2-D DCT and IDCT Processors Based on CORDIC Rotation	Tze-Yun Sung	560-118
Memory-Efficient and High-Speed Line-Based Architecture for 2-D Discrete Wavelet Transform with Lifting Scheme	Tze-Yun Sung	560-119
Suboptimal MIMO Detector based on Viterbi Algorithm	Jin Lee and Sin-Chong Park	560-463
Frequency domain constructed redundant bases for denoising	Jianyu Lin and Graham Goodwin	560-195
Spectrum flipping for wavelet packet decomposition	Jianyu Lin	560-194

Monday, April 16, 2007

PLENARY LECTURE 3

Intentional 3D Vision in Robotics

Professor S. Y. Chen

University of Hamburg

Zhejiang University of Technology

CHINA

Abstract: The 3D model of an object or environment is very important to many practical applications, e.g. hypermedia techniques, interactive handling/working, virtualized reality, interactive exploration, robot autonomous applications, etc. An efficient way to obtain such models is using 3D vision. However, since no prior information about the targets is available before the vision task is taking place, it is desired that the robot is capable to purposively decide how to finish the modeling task without human interference. The project investigates the issue of intentional perception planning in which there are three related problems to be tackled, i.e. perception planning, purposive sensor configuration, and 3D model construction. They are considered in an autonomous vision system which employs an observing strategy for target modeling. The research develops methods to actively decide the viewpoints and sensor configuration to observe specific aspects of the target and estimates the modeling state or completeness. The system analyzes the obtained information and performs the perception strategy that automatically generates a multi-view plan. The research would lead to fully automate the modeling process and have the mechanism to autonomously move and change sensing parameters for complete 3D data recovery.

SESSION: Advances on Image Processing I

Chair: Li Jing, Ching-Tang Hsieh

A Relevance Feedback Image Retrieval Scheme using Combination of Color and Shape Features	Yuexiang Shi, Donghui Zhu, Li Wen	560-103
Robust Zero-watermarking Scheme Using Local Invariant Keypoints	Li Jing, Shuhong Li	560-172
Principal Component Analysis for Minutiae Verification on Fingerprint Image	Ching-Tang Hsieh, Shys-Rong Shyu	560-122
Low-Power and High-Performance 2-D DWT and IDWT Architectures Based on 4-tap Daubechies Filter	Tze-Yun Sung	560-128
Additive Texture Information Extraction Using Color Coherence Vector	Ki-Hyun Kang, Yong-In Yoon, Jong-Soo Choi, Jin-Tae Kim, Hasung Koo, Jong-Ho Choi	560-154
An Image Encryption Algorithm Based on Multi-Dimensional Orthogonal Sequence	Shuhong Li, Li Jing, Xing Gao	560-149

SESSION: Advances on Image Processing II

Chair: Ho-Keun Song, Feng Li

A robust corresponding points for uncalibrated stereo images	Yong-In Yoon, Ki-Hyun Kang, Jong-Soo Choi, Ho-Keun Song, Jong-Ho Choi	560-284
The Lateral Restraint Network Model on the Processing of Image	Li Feng, Zhao Shimin, Liu Xiaoqiang	560-325
An Adaptive Wavelet-Based Watermarking Using Shift-Orthogonal Finite-Length Sequences	Shuhong Li, Qiaorong Zhang, Yadong Zhang	560-212
An Efficient Rearrangement of Wavelet Packet Coefficients for Embedded Quad-Tree Image Coding	Tze-Yun Sung, Hsi-Chin Hsin	560-308
Automatic Sightline Stabilisation in Noisy Imagery	Shangqi Bao, Jason F. Ralph	560-286
DLDA-based iris recognition from image sequences with various focus information	Byungjun Son, Sung-hyuk Cha, Yillbyung Lee	560-443
Image Reconstruction Using a 2D M-channel Perfect Reconstruction Filter Bank with an Optimized Adaptive Interpolation Kernel	J. Y. Kim, K. J. Kim and S. W. Nam	560-437

Tuesday, April 17, 2007

SESSION: Advances on Multimedia & Signal Processing

Chair: Tze-Yun Sung, Shinfeng Lin

Numerical Accuracy and Hardware Trade-Offs for Fixed-Point CORDIC Processor for Digital Signal Processing System	Tze-Yun Sung	560-120
Error Resilience using a Reversible Data Embedding Technique in H.264/AVC	Shinfeng D. Lin, Yu-Lung Su and Jhih-Yu Huang	560-405
Flexible jacket matrices for cooperative multi-agent network	Moon Ho Lee, Xueqin Jiang, Zhu Chen, Chang-hui Choe	560-483
An improvement on the detection of skin cancer based on raman spectroscopy	Aaron Park, In-Wook Jung, Seong-Joon Baek, Jin Young Kim, Seung You Na	560-424
Applying recurrence quantification and spectral analysis to represent nasalization in speech signals	Yousif A. El-Imam, Ahmad S. Elwakil	560-099
Applied variable step size algorithm to dual-adaptive noise canceller	Yen-Nien Wang, Hui-Hsiung Liao, Tsai-Sheng Lee	560-206
New Figures Of Merit For Range Resolution Radar Using Hamming And Euclidean Distance Concepts	K. Raja Rajeswari, P. Srihari, P. Rajesh Kumar, M. Murali, V. Jagan Naveen, G. Manmadha Rao	567-182
Fuzzy Modelling For Discrimination And Merit Factor Of Radar Signals For Range Resolution	P. Sri Hari, P. Rajesh Kumar, D. Tirumal Rao, K. Raja Rajeswari	567-183
A New Sidelobe Reduction Technique For Range Resolution Radar	K. Raja Rajeswari, M. Uttara Kumari, Ch. Srinivas, B. Leela Prakash, K. Srihari Rao	567-184