Confocal imaging system is a newly developed technique in optical metrology and widely applied to profile inspection in the filed of industrial components and bioscience. With industrial manufacturing fast development...
详细信息
ISBN:
(纸本)9781934272350
Confocal imaging system is a newly developed technique in optical metrology and widely applied to profile inspection in the filed of industrial components and bioscience. With industrial manufacturing fast development, the requirement of high resolution and fast 3D profile inspection is increased. The research results are satisfied in the effort of high resolution study and reached nanometer scale. However, the 3D profile inspection speed still could not satisfy the large integrated manufacturing requirement. On-line inspection and fast sampling in large integrated manufacturing is very important. Some research effort for parallel confocal imaging system was done to improve fast 3D profile inspection. However, confocal image processing speed is relatively slow in principle. This paper presents a fast algorithm to improve confocal image processing speed. With a new configuration of fiber confocal image system, a continuous experiment was done for the study of system repeatability and invariance of the depth response. A three-point interpolation method is used in confocal image processing. The experimental result using fast algorithm is satisfied comparing with the conventional confocal image sampling method and theoretical calculation.
The paper suggests new fast algorithm for approximate Hoteling basis construction which is known to be important in pattern recognition. In contrary to traditional methods the algorithm permits to classify signals wit...
详细信息
ISBN:
(纸本)0819429155
The paper suggests new fast algorithm for approximate Hoteling basis construction which is known to be important in pattern recognition. In contrary to traditional methods the algorithm permits to classify signals with high capacity. The presented method is based on wavelet transform and it is fast. Efficiency of the constructed approximate Hoteling filter is demonstrated. Experimental results are obtained for texture classification problem. The comparison with recognition using signs obtained via approximate Karhunen-Loeve transform is conducted.
A fast algorithm for matrix embedding steganography is proposed in this paper. Matrix embedding encodes the cover image and the secret message with an error correction code and modifies the cover image according to th...
详细信息
A fast algorithm for matrix embedding steganography is proposed in this paper. Matrix embedding encodes the cover image and the secret message with an error correction code and modifies the cover image according to the coding result. The modification to the cover image is the coset leader of the error correction code, and it is computationally complex to find the coset leader. This paper proposes a fast algorithm to find the coset leader by using a lookup table algorithm. The proposed. algorithm is suitable for matrix embedding steganography using Hamming code and random linear code. In our scheme, the syndrome of the coset is used to search for the coset leader in the standard array of the error correction code. For the Hamming code, we improved the parity check matrix of the code in order to make the syndrome indicate the coset leader by itself. Therefore, it is not necessary to search for the coset leader in a table. For the random linear code, this method is effective for most cosets, and we only memorize the coset leaders that cannot be identified by their syndromes. With this approach, the size of the table can be reduced significantly, and the computational complexity of embedding can be decreased. The proposed fast embedding algorithm has the same embedding efficiency as the conventional matrix embedding. Compared with the existing fast matrix embedding algorithms, the computational complexity of the proposed scheme is decreased significantly for the steganographic systems with low and medium embedding rates. (C)2013 Elsevier Inc. All rights reserved.
A neural classifier whose training can be executed very effectively is proposed to overcome the disadvantages of the method of potential functions. The power of the method of potential functions is limited by the seve...
详细信息
A neural classifier whose training can be executed very effectively is proposed to overcome the disadvantages of the method of potential functions. The power of the method of potential functions is limited by the severe requirements for computation time and storage. After mapping the computational structure of the method of potential functions to a three-layered feedforward network, a new fast learning algorithm is applied to train the net. By the characteristics of the new training algorithm, the network's complexity can be largely reduced without degrading its performance too much. That is to say, the algorithm will enable the network to learn in a very fast and, moreover, very effective manner.
This paper presents a new algorithm for the fast computation of multidimensional (m-D) discrete cosine transform (DCT) with size N-1 x N-2 x (...) x N-m, where N-i is a power of 2 and N-1 less than or equal to 256, by...
详细信息
This paper presents a new algorithm for the fast computation of multidimensional (m-D) discrete cosine transform (DCT) with size N-1 x N-2 x (...) x N-m, where N-i is a power of 2 and N-1 less than or equal to 256, by using the tensor product decomposition of the transform matrix. It is shown that the m-D DCT or inverse discrete cosine transform (IDCT) on these small sizes can be computed using only one-dimensional (1-D) DCTs and additions and shifts. If all the dimensional sizes are the same, the total number of multiplications required for the algorithm is only 1/m times of that required for the conventional row-column method. We also introduce approaches for computing scaled DCTs in which the number of-multiplications is considerably reduced.
In some cases such as heating objects by microwave power, variation of the temperature in objects changes their dielectric permittivity and then leads to variation of the electromagnetic (EM) field. The commonly adopt...
详细信息
In some cases such as heating objects by microwave power, variation of the temperature in objects changes their dielectric permittivity and then leads to variation of the electromagnetic (EM) field. The commonly adopted method of simulation for the heating procedure is to repeatedly execute the full-wave simulation for each small time intervals, which results in a heavy computation burden. A fast algorithm for simulating 2-D sceneries is proposed for the first time. It starts with a full-wave simulation to get an initial EM field, and then corrects the EM field in an efficient way rather than repeating the full-wave simulation when the object's permittivity has slightly changed. A few sample numerical results are presented and analyzed, which verifies the accuracy, robustness, and effectiveness of the proposed fast algorithm.
We present a novel fast algorithm for flow simulations using the discrete vortex method, DVM, for problems with periodic boundary conditions. In the DVM, the solution of the velocity field induced by interactions amon...
详细信息
We present a novel fast algorithm for flow simulations using the discrete vortex method, DVM, for problems with periodic boundary conditions. In the DVM, the solution of the velocity field induced by interactions among N discrete vortex particles is governed by the Biot-Savart law and, therefore, leads to a computational cost proportional to O(). The proposed algorithm combines exponential and power series expansions implemented using a divide and conquer strategy to accelerate the calculation of the cotangent kernel that models periodic boundary conditions. The fast multipole method, FMM, is applied for the solution of singular terms appearing in the power series expansion and also for the exponential series expansion. Error and computational cost analyses are performed for the individual steps of the algorithm for double and quadruple machine precision. The current method presents more accurate solutions when compared to those obtained by periodic domain replication using the free-field FMM kernel. The novel algorithm provides computational savings of nearly 240 times for double-precision simulations with one million particles when compared to the direct calculation of the Biot-Savart law.
Background: There is an increasing need to routinely and quickly compare multiple sequences of, for example, bird flu virus genomes to infer their evolutionary relationship. This entails a fast simultaneous inference ...
详细信息
Background: There is an increasing need to routinely and quickly compare multiple sequences of, for example, bird flu virus genomes to infer their evolutionary relationship. This entails a fast simultaneous inference of both sequence alignment and phylogeny. Current methods cannot meet the speed requirement though a high phylogeny accuracy is maintained in such scenarios. Objective: We propose a fast algorithm for constructing Multiple sequence Alignment and Phylogeny (FAMAP) from closely related DNA sequences. Method: FAMAP is essentially a sequentially-inputting algorithm and can be implemented in a progressive fashion, i.e., adding a new sequence into an existing tree or multiple sequence alignment. Its time complexity is O[NP(L)] + O(NG) and its space complexity is O(N) + O(G) + O[Q(L)] , where N is the number of sequences, N is the number of mutations on the phylogeny, L is the maximum length of the sequences, and P(L) and Q(L) are the time and space complexity of aligning a pair of sequences of length L, depending on the pairwise alignment algorithm employed. Results: Intensive simulation studies shows that our method is superior in terms of speed over other popular methods and has comparable accuracy of both multiple sequence alignment and the phylogeny. Conclusion: Our new algorithm might be one of the best choices when the user wants to quickly obtain a reliable phylogeny estimation from dozens of closely related long sequences.
The fast multipole method (FMM) is used to solve the electromagnetic scattering from two-dimensional conducting bodies of arbitrary shape when a iterative method is used to solve the electric field integral equation (...
详细信息
ISBN:
(纸本)962442117X
The fast multipole method (FMM) is used to solve the electromagnetic scattering from two-dimensional conducting bodies of arbitrary shape when a iterative method is used to solve the electric field integral equation (EFIE). FMM reduces the complexity of computing the matrix-vector multiplication from O(N-2)to O(N-15) Instead of the conjugate gradient (CG) method, in this paper the biconjugate gradient (BICG) method is used to accelerate the iteration process.
Low-efficiency of current algorithm for structural frequency domain response subjected to complex distributed random load, leads to the inconvenience of engineering application. New fast algorithm is introduced to sol...
详细信息
ISBN:
(纸本)9781846260667
Low-efficiency of current algorithm for structural frequency domain response subjected to complex distributed random load, leads to the inconvenience of engineering application. New fast algorithm is introduced to solver above problems in the article. Firstly, the cross power spectrum function of distributed random load is expanded by two-dimension Legendre polynomial, and the infinite distributing information is expressed by finite parameters. Then the dynamic response computation of distributed random load is transformed into that of distributed harmonic load. The response power spectrum for each frequency point can be obtained by vector multiplication. Compared the fast algorithm with regular algorithm, it can be concluded that the fast algorithm is accurate, and its efficiency is much higher than regular algorithm. Moreover, it is convenient to consider the coherences of distributed random load. Finally, with software of finite element analysis, computation software included the fast algorithm is made for computation of distribution random load.
暂无评论