Browsing by Author "Hsu, Chi-Yao"

Now showing 1 - 5 of 5

Accurate and rapid alignment of laser scanned 3D surface using TSK-type neural-fuzzy network-based coarse-to-fine strategy
(2012-10-01)
Chang, Jyun-Wei
;
Lin, Sheng-Fuu
;
Hsu, Chi-Yao
;
電機工程學系
;
Department of Electrical and Computer Engineering
Aligning a laser scanned three-dimensional (3D) surface is considered a critical step in object recognition, shape analysis, and automatic visual inspection. Two major concerns for the alignment task are execution time and alignment accuracy. Recently, neural network-based methods have become very popular due to their high efficiency. However, such methods experience difficulty in reaching high accuracy because the use of principal component analysis (PCA) to perform coarse alignment causes a large alignment error. Thus, a TSK-type neural-fuzzy network (TNFN)-based coarse-to-fine 3D surface alignment scheme is proposed in the current paper. Compared with traditional neural network-based approaches, the proposed method provides a coarse-to-fine alignment approach to ensure the accurate pose estimated by TNFN in the coarse phase, as well the high alignment speed provided by TNFN-based surface modeling in the fine phase. Experimental results demonstrate the superior performance of the proposed 3D surface alignment system over existing systems. (C) 2012 Elsevier Ltd. All rights reserved.
AN EFFICIENT 3D MODEL RETRIEVAL BASED ON PRINCIPAL AXES ANALYSIS AND FEATURE INTEGRATION
(2011-06-01)
Lin, Sheng-Fuu
;
Wu, Chin-Chia
;
Hsu, Chi-Yao
;
Hsu, Dou-Chih
;
電機工程學系
Three-dimensional (3D) model retrieval has gathered great importance in recent years, since the number of available 3D models on the Internet has drastically increased. Many content-based 3D model retrieval approaches have been proposed. Among these methods, visual similarity-based methods have shown higher retrieval accuracy. However, because these methods capture enormous shape features from different viewpoints or locations, a large amount of calculation and comparison is necessary. Furthermore, there is a trade-off between retrieval accuracy and speed. In this paper, a 3D model retrieval method constituting Continuous Principal Component Analysis (CPCA), Fourier descriptors, and Zernike moments is proposed. CPCA is applied to extract significant shape features based on projecting the model along the principal axes. Then, Fourier descriptors and Zernike moments are used to provide shape descriptors with rotation invariants. In addition, a feature integration process combines them. A strategy of similarity measure is proposed to solve the axes switching problem. To conclude, the experimental results show that the approach outperforms SECTORS(2) and D2,(18) and has slightly better retrieval results than Light Field Descriptor (LFD)(6) and spin-image signatures.(3) Moreover, the approach is more efficient and the storage size is much less.
Data mining-based hierarchical cooperative coevolutionary algorithm for TSK-type neuro-fuzzy networks design
(2013-08-01)
Hsu, Chi-Yao
;
Lin, Sheng-Fuu
;
Chang, Jyun-Wei
;
電機工程學系
;
Department of Electrical and Computer Engineering
This study proposes a data mining-based hierarchical cooperative coevolutionary algorithm (DMHCCA) for TSK-type neuro-fuzzy networks design. The proposed DMHCCA consists of two-level evolutions: the neuro-level evolution (NULE) and the network-level evolution (NWLE). In NULE, a data mining-based evolutionary learning algorithm is utilized to evolve neurons. The good combinations of neurons evolved in NULE are reserved for being the initial populations of NWLE. In NWLE, the initial population are mated and mutated to produce new structure of networks. Similar to NULE, the good neurons of evolved network in NWLE are inserted into the NULE. Thus, by interactive two-level evolutions, the neurons and structure of network can be evolved locally and globally, respectively. Simulation results using DMHCCA are reported and compared with other existing models. Application of DMHCCA to a three-dimensional (3D) surface alignment task is also described, and experimental results are presented better performance than other alignment systems.
Precise image alignment using cooperative neural-fuzzy networks with association rule mining-based evolutionary learning algorithm
(2012-02-01)
Hsu, Chi-Yao
;
Cheng, Yi-Chang
;
Lin, Sheng-Fuu
;
電機工程學系
;
Department of Electrical and Computer Engineering
Precise image alignment is considered a critical issue in industrial visual inspection, since it performs an accurate pose to the object in inspected images. Recently, image alignment based on neural networks has become very popular due to its performance at speed. However, such a method has difficulty when applied to the alignment of images on a large range of affine transformation. To address this, a cooperative neural-fuzzy network (CNFN) with association rule mining-based evolutionary learning algorithm (ARMELA) is proposed. Unlike traditional neural network-based approaches, the proposed CNFN utilizes a coarse-to-fine alignment procedure to adapt image alignment to a larger range of affine transformation. The proposed ARMELA combines the self-adaptive method and association rules selection method to self-adjust the structure and parameters of the neural-fuzzy network. Furthermore, L2 regularization is adopted to control ARMELA such that the convergence speed increases. Experimental results show that the performance of the proposed scheme is superior to the traditional neural network methods in terms of accuracy and robustness. (c) 2012 Society of Photo-Optical Instrumentation Engineers (SPIE). [DOI: 10.1117/1.OE.51.2.027006]
Reinforcement evolutionary learning using data mining algorithm with TSK-type fuzzy controllers
(2011-04-01)
Hsu, Chi-Yao
;
Hsu, Yung-Chi
;
Lin, Sheng-Fuu
;
電機工程學系
;
Department of Electrical and Computer Engineering
Reinforcement evolutionary learning using data mining algorithm (R-ELDMA) with a TSK-type fuzzy controller (TFC) for solving reinforcement control problems is proposed in this study. R-ELDMA aims to determine suitable rules in a TFC and identify suitable and unsuitable groups for chromosome selection. To this end, the proposed R-ELDMA entails both structure and parameter learning. In structure learning, the proposed R-ELDMA adopts our previous research - the self-adaptive method (SAM) - to determine the suitability of TFC models with different fuzzy rules. In parameter learning, the data-mining based selection strategy (DSS), which proposes association rules, is used. More specifically, DSS not only determines suitable groups for chromosomes selection but also identifies unsuitable groups to be avoided selecting chromosomes to construct a TFC. Illustrative examples are presented to show the performance and applicability of the proposed R-ELDMA. Crown Copyright (C) 2010 Published by Elsevier B. V. All rights reserved.