A new blind restoration method applying Real-coded genetic algorithm (RcGA) will be proposed, and this method will be proven valid for the blurred image restoration with unidentified degradation in the experiments. In this restoration method, the degraded and blurred image is going to get restricted to the images possible to be expressed in the point spread function (PSF), then the restoration filter for this degraded image, which is also the 2-dimentional inverse filter, will be searched among several points applying RcGA. The method will enable to seek efficiently among vast solution space consists of numeral coefficient filters. And perceiving the essential features of the spectrum in the frequency space, an evaluation function will be proposed. Also, it will be proposed to apply the Rolling-ball transform succeeding an appropriate Gaussian degrade function against the dual degraded image with blur convoluting impulse noise. By above stated features of this restoration method, it will enable to restore the degraded image closer to the original within a practical processing time. Computer simulations verify this method for image restoration problem when the factors causing image distortions are not identified.

A simple system configuration was used to generate transform-limited optical pulses at 160 Gbit/s in the sub-picosecond range (625 fs). Pulse compression was achieved by broadening the spectrum using supercontinuum generation followed by a linear frequency chirping compensation.

Fuzzy value is a fuzzy set on interval [0,1], whose α-cuts are all closed intervals for α [0,1], i.e., fuzzy value is a fuzzy number on [0,1]. In this note, we introduce three kinds of metrics di (i=1,2,3) into fuzzy-valued space m[0,1] and consider continuity of fuzzy-valued operations on metric spaces (m[0,1], di) (i=1,2,3). The obtained results will provide some theoretical bases for numeral calculation of fuzzy-valued operations.

Since the deployment of base stations (BS's) is far from optimum in 3-dimensional (3-D) space, i.e., the vertical baseline is relatively shorter than the planar baseline, the geometric degradation of precision of the altitude estimate is larger than that of the planar location. This paper considers the problem of 3-D range location and attempt to improve the altitude estimate. We first use a volume formula of tetrahedron to transform the range measurements to the volume measurements, then a novel pseudo-linear solution is proposed based on a linear relationship between the rectangular and the volume coordinates. Theory analysis and numerical examples are included to show the improved accuracy of the altitude estimate of mobile location. Finally, an improved estimate of 3-D mobile location is given by solving a set of augmented linear equations.

In this paper, a high performance 3232-bit multiplier for a DSP core is proposed. The multiplier is composed of a block of Booth Encoder, a block of data compression, and a block of a 64-bit adder. In the block of Booth encoder, a conditional sign decision Booth encoder that reduces the gate delay and power consumption is proposed. In the block of data compression, 4-2 and 9-2 data compressors based on a novel compound logic are used for the efficient compressing of extra sign bit. In the block of 64-bit adder, an adaptive MUX-based conditional select adder with a separated carry generation block is proposed. The proposed 3232-bit multiplier is designed by a full-custom method and there are about 28,000 transistors in an active area of 900 µm 500 µm with 0.25 µm CMOS technology. From the experimental results, the multiplication time of the multiplier is about 3.2 ns at 2.5 V power supply, and it consumes about 50 mW at 100 MHz.

A novel method for the primitive description of the multiprimitive texture is proposed. This method segments a texture by the watershed algorithm into fragments each of which contains one grain. The similar fragments are grouped by the cluster analysis in the feature space whose basis is the morphological size density. Each primitive is extracted as the grain of the central fragment in each cluster.

The starting point of this paper is the definition of a nonlinear model of the read out process in high density optical discs. Under high density condition, the signal read out is not a linear process, and suffers also from cross talk. To cope with these problems, the identification of a suitable nonlinear model is required. A physical model based on the optical scalar theory is used to identify the kernels of a nonlinear model based on the Volterra series. Both analysis and simulations show that a second order bidimensional model accurately describes the read out process. Once equipped with the Volterra channel model, we evaluate the performance of various nonlinear receivers. First we consider Nonlinear Adaptive Volterra Equalization (NAVE). Simulations show that the performance of classical structures for linear channels is significantly affected by the nonlinear response. The nonlinear NAVE receiver can achieve better performance than Maximum Likelihood Sequence Estimator (MLSE), with lower complexity. An innovative Nonlinear Maximum Likelihood Sequence Estimator (NMLSE), based on the combination of MLSE and nonlinear Inter-Symbol Interference (ISI) cancellation, is presented. NMLSE offers significant advantages with respect to traditional MLSE, and performs better than traditional equalization for nonlinear channels (like NAVE). Finally, the paper deals with cancellation of cross talk from adjacent tracks. We propose and analyze an adaptive nonlinear cross talk canceller based on a three spot detection system. For the sake of simplicity, all the performance comparisons presented in this paper are based on the assumption that noise is Additive, White, and Gaussian (AWGN model).

The lossy magnetic composite material made from soft magnetic metal powder and rubber is widely used as an EMI countermeasure material, due to its higher magnetic loss than those of spinel ferrites in microwave frequencies. In this paper, we clarify the material characteristics by measuring the relative complex permeability r and relative complex permittivity r of two kinds of composite materials in microwave frequencies. Since the composite materials are anisotropic, both r and r are measured as diagonal tensors by utilizing extended S-parameter method. The results show that the imaginary part of r of flaky-powder composite exceeded the Snoek's limit for the spinel ferrites which has been reported so far. The measured r and r are partially compared with those measured by cavity resonator method, and good agreement is obtained.

This letter presents a new approach for obtaining the expected waiting time for packets under the drop-head (also called a drop-from-front) scheme for buffer management. The results show that the drop-head scheme is more effective in reducing queueing delays than the drop-tail scheme.

There are several attempts to generate chaotic binary sequences by using one-dimensional maps. From the standpoint of engineering applications, it is necessary to evaluate statistical properties of sample sequences of finite length. In this paper we attempt to evaluate the statistics of chaotic binary sequences of finite length. The large deviation theory for dynamical systems is useful for investigating this problem.

In this paper we present the analog architecture and the implementation of an on-chip learning Multi Layer Perceptron network. The learning algorithm is based on Back Propagation but it exhibits increased capabilities due to local learning rate management. A prototype chip (SLANP, Self-Learning Neural Processor) has been designed and fabricated in a CMOS 0.7 µm minimum channel length technology. We report the experimental results that confirm the functionality of the chip and the soundness of the approach. The SLANP performance compare favourably with those reported in the literature.

The radar cross section (RCS) of a dielectric-coated cylindrical cavity was measured and the measurements were compared with those calculated according to the iterative physical optics (IPO). The IPO analysis used the equivalent-impedance boundary condition (EIBC) based on transmission-line theory which takes into account the thickness of the coating. It was consequently found that this condition is much more effective than the ordinary-impedance boundary condition based on the intrinsic impedance of the material.

The cooperative and competitive network suitable for circuit realization is presented, based on the network proposed by Amari and Arbib. To ensure WTA process, the output function of the original network is replaced with the piecewise linear function and supplying the inputs as pulse waveforms is obtained. In the SPICE simulations, it is confirmed that the network constructed by operational amplifiers attains WTA process, even if the scale of the network becomes large.

Algorithms are presented for the four elementary arithmetic operations, to perform reliable floating-point arithmetic operations. These arithmetic operations can be achieved by applying residue techniques to the weighted number systems and performed with no accuracy lost in the process of the computing. The arithmetic operations presented can be used as elementary tools (on many existing architectures) to ensure the reliability of numerical computations. Simulation results especially for the solutions of ill-conditioned problems are given with emphasis on the practical usability of the tools.

This paper investigates the behavior of one-dimensional discrete-time binary cellular neural networks with both the A- and B-templates and gives the necessary and sufficient conditions for the above network to be stable for unspecified fixed boundaries.

This paper presents a new Reinforcement Learning (RL) method, called "Labeling Q-learning (LQ-learning)," to solve the partially obervable Markov Decision Process (POMDP) problems. Recently, hierarchical RL methods are widely studied. However, they have the drawback that the learning time and memory are exhausted only for keeping the hierarchical structure, though they wouldn't be necessary. On the other hand, our LQ-learning has no hierarchical structure, but adopts a new type of internal memory mechanism. Namely, in the LQ-learning, the agent percepts the current state by pair of observation and its label, and then, the agent can distinguish states, which look as same, but obviously different, more exactly. So to speak, at each step t, we define a new type of perception of its environment õt=(ot,θt), where ot is conventional observation, and θt is the label attached to the observation ot. Then the classical RL-algorithm is used as if the pair (ot,θt) serves as a Markov state. This labeling is carried out by a Boolean variable, called "CHANGE," and a hash-like or mod function, called Labeling Function (LF). In order to demonstrate the efficiency of LQ-learning, we will apply it to "maze problems" in Grid-Worlds, used in many literatures as POMDP simulated environments. By using the LQ-learning, we can solve the maze problems without initial knowledge of environments.

In this paper, we present the following schemes for a content-based image search: (1) A fast image search algorithm that can significantly reduce similarity calculation compared to a full comparison of every database image. (2) A compact image representation scheme that can describe the global/local information of the images and provide successful retrieval performance. For fast searches, a tree is constructed by successfully dividing nodes into the desired depth level by working from the root to the leaf nodes using the k-means algorithm. When the query is completed, we traverse the tree top-down by minimizing the route taken between the query image and node centroid until we meet the undivided nodes. Within undivided nodes, the algorithm of triangle inequality is used to find the images most similar to the query. For compact image representation, RGB color histogram features which are quantized into 16 bins each of the R, G, and B channels are used for global information. Dominant hue, saturation, and value which are extracted from the HSV joint histogram in the localized regions within the image are used for local information. These features are sufficiently compact to index image features in large database systems. For experiments on the retrieval efficiency, the use of the proposed method provided substantial performance benefits by reducing the image similarity calculation up to an average of a 96% and for experiments on the retrieval effectiveness, in the best case, it provide a 36.8% recall rate for a whale query image and a 100% precision rate for an eagle query image. The overall performance was a 20.0% recall rate and a 72.5% precision rate.

Let us introduce n ( 2) mappings fi (i=1,2,,n) defined on complete linear metric spaces (Xi-1, ρ) (i=1,2,,n), respectively, and let fi:Xi-1 Xi be completely continuous on bounded convex closed subsets Xi-1(0) Xi-1, (i=1,2,,n 0), such that fi(Xi-1(0)) Xi(0). Moreover, let us introduce n set-valued mappings Fi : Xi-1 Xi (Xi)(the family of all non-empty closed compact subsets of Xi), (i=1,2,,n 0). Here, we have a fixed point theorem on the successively recurrent system of set-valued mapping equations: xi Fi(xi-1, fi(xi-1)), (i=1,2,,n 0). This theorem can be applied immediately to analysis of the availability of system of circular networks of channels undergone by uncertain fluctuations and to evaluation of the tolerability of behaviors of those systems. In this paper, mathematical situation and detailed proof are discussed, about this theorem.

Cellular Neural Networks (CNNs) have been developed as a high-speed parallel signal-processing platform. In this paper, a generalized two-layer cellular neural network model is proposed for image processing, in which two templates are introduced between the two layers. We found from the simulations that the two-layer CNNs efficiently behave compared to the single-layer CNNs for the many applications of image processing. For examples, simulation problems such as linearly non-separable task--logic XOR, center point detection and object separation, etc. can be efficiently solved with the two-layer CNNs. The stability problems of the two-layer CNNs with symmetric and/or special coupling templates are also discussed based on the Lyapunov function technique. Its equilibrium points are found from the trajectories in a phase plane, whose results agree with those from simulations.

This paper describes a hybrid force-directed self-organizing neural network approach to printed circuit board (PCB) placement with consideration of electromagnetic compatibility (EMC). In most of the conventional PCB automatic placement algorithms, the only factor considered in the objective function is minimized total net length. However, for today's high speed and high density PCB, EMC compliance cannot be met by such single objective. To tackle this problem, the presented algorithm takes EMC into consideration, besides component overlap and minimized total net length. These factors are optimized by means of an adapted self-organizing map. Comparison of simulated placement results as well as actual measurements with commercial softwares confirms the effectiveness of the proposed method.