Cellular Neural Network Paradigm Research Articles

A CNN-based approach for a class of non-standard hyperbolic partial differential equations modeling distributed parameters (nonlinear) control systems

The present paper considers a systematic approach within the framework of Neural Mathematics for constructing a computational procedure. This procedure aims to solve a class of problems arising from the control of the systems with distributed parameters; these systems are modeled by second-order one-dimensional hyperbolic partial differential equations (hPDEs) with non-standard boundary conditions. The procedure reveals an explicit algorithmic parallelism and is mainly based on the combination of two powerful “tools”: a convergent Method of Lines (MoL) and the Cellular Neural Network (CNN) paradigm. The role of the Courant-Isaacson-Rees rule and of the Riemann invariants for a correct application of the MoL is emphasized. The procedure is illustrated on a control engineering application – the overhead crane with flexible cable – within a more general context which includes modeling based on the generalized Hamilton variational principle, synthesis of a stabilizing controller via the Control Lyapunov Functional (CLF), qualitative analysis, numerical solving using the proposed computational procedure, numerical simulations and the evaluation of the performances for the closed loop system. The procedure ensures the convergence of the approximation, preserves the basic properties and the Lyapunov stability of the solution of the initial problem and reduces the systematic errors.

Simulation, Modeling, and Analysis of Soliton Waves Interaction and Propagation in CNN Transmission Lines for Innovative Data Communication and Processing

We present an innovative approach to study the interaction between oblique solitons, using nonlinear transmission lines, based on Cellular Neural Network (CNN) paradigm. A single transmission line consists of a 1D array of cells that interact with neighboring cells, through both linear and nonlinear connections. Each cell is controlled by a nonlinear Ordinary Differential Equation, in particular the Korteweg de Vries equation, which defines the cell status and behavior. Two typologies of CNN transmission lines are modelled: crisscross and ring lines. In order to solve KdV equations two different methods are used: 4th-order Runge-Kutta and Forward Euler methods. This is done to evaluate their accuracy and stability with the purpose of implementing CNN transmission lines on embedded systems such as FPGA and microcontrollers. Simulation/analysis Graphic User Interface platforms are designed to conduct numerical simulations and to display elaboration results. From this analysis it is possible both to identify the presence and the propagation of soliton waves on the transmission lines and to highlight the interaction between solitons and rich nonlinear dynamics. With this approach it is possible to simulate and develop the transmission and processing of information within large brain networks and high density sensor systems.

FPGA-based distributed computing microarchitecture for complex physical dynamics investigation.

In this paper, we present a distributed computing system, called DCMARK, aimed at solving partial differential equations at the basis of many investigation fields, such as solid state physics, nuclear physics, and plasma physics. This distributed architecture is based on the cellular neural network paradigm, which allows us to divide the differential equation system solving into many parallel integration operations to be executed by a custom multiprocessor system. We push the number of processors to the limit of one processor for each equation. In order to test the present idea, we choose to implement DCMARK on a single FPGA, designing the single processor in order to minimize its hardware requirements and to obtain a large number of easily interconnected processors. This approach is particularly suited to study the properties of 1-, 2- and 3-D locally interconnected dynamical systems. In order to test the computing platform, we implement a 200 cells, Korteweg-de Vries (KdV) equation solver and perform a comparison between simulations conducted on a high performance PC and on our system. Since our distributed architecture takes a constant computing time to solve the equation system, independently of the number of dynamical elements (cells) of the CNN array, it allows us to reduce the elaboration time more than other similar systems in the literature. To ensure a high level of reconfigurability, we design a compact system on programmable chip managed by a softcore processor, which controls the fast data/control communication between our system and a PC Host. An intuitively graphical user interface allows us to change the calculation parameters and plot the results.

An improved fuzzy cellular neural network (IFCNN) for an edge detection based on parallel RK(5,6) approach

Parallel RK(5,6) algorithms are employed to validate the potential behaviour under improved fuzzy cellular neural network paradigm for an edge detection problem is the novelty of this paper. The function of the simulator is capable of performing raster fuzzy cellular neural network simulation for any kind as well as any size of input image. The raster pseudo code exploits the latency properties of improved fuzzy cellular neural network coupled with different parallel numerical integration algorithms. Results provide a better understanding about single layer/raster behaviour simulation.

ANALOG COMPUTING FOR A NEW NUCLEAR REACTOR DYNAMIC MODEL BASED ON A TIME-DEPENDENT SECOND ORDER FORM OF THE NEUTRON TRANSPORT EQUATION

This paper considers the concept of analog computing based on a cellular neural network (CNN) paradigm to simulate nuclear reactor dynamics using a time-dependent second order form of the neutron transport equation. Instead of solving nuclear reactor dynamic equations numerically, which is time-consuming and suffers from such weaknesses as vulnerability to transient phenomena, accumulation of round-off errors and floating-point overflows, use is made of a new method based on a cellular neural network. The state-of-the-art shows the CNN as being an alternative solution to the conventional numerical computation method. Indeed CNN is an analog computing paradigm that performs ultra-fast calculations and provides accurate results. In this study use is made of the CNN model to simulate the space-time response of scalar flux distribution in steady state and transient conditions. The CNN model also is used to simulate step perturbation in the core. The accuracy and capability of the CNN model are examined in 2D Cartesian geometry for two fixed source problems, a mini-BWR assembly, and a TWIGL Seed/Blanket problem. We also use the CNN model concurrently for a typical small PWR assembly to simulate the effect of temperature feedback, poisons, and control rods on the scalar flux distribution.

Implementation of a cellular neural network–based segmentation algorithm on the bio-inspired vision system

Based on the cellular neural network (CNN) paradigm, the bio-inspired (bi-i) cellular vision system is a computing platform consisting of state-of-the-art sensing, cellular sensing-processing and digital signal processing. This paper presents the implementation of a novel CNN-based segmentation algorithm onto the bi-i system. The experimental results, carried out for different benchmark video sequences, highlight the feasibility of the approach, which provides a frame rate of about 26 frame/sec. Comparisons with existing CNN-based methods show that, even though these methods are from two to six times faster than the proposed one, the conceived approach is more accurate and, consequently, represents a satisfying trade-off between real-time requirements and accuracy.

On Analog Networks and Mixed-Domain Spatio-Temporal Frequency Response

In this paper, we extend the linear cellular neural network (CNN) paradigm by introducing temporal derivative diffusion connections between neighboring cells. Our proposal results in an analog network topology for implementing general continuous-time discrete-space mixed-domain 3-D rational transfer functions for linear filtering. The network connections correspond one-to-one to the transfer function coefficients. The mixed-domain frequency response is treated as a temporal frequency-dependent spatial function and we show how nonseparable properties of the spatio-temporal magnitude response can be derived from the combination of: 1) sinusoidal functions of spatial frequencies and 2) polynomials of the continuous-time frequency in the 3-D frequency response expression. A generic VLSI-compatible implementation of the network based on continuous-time integrators is also proposed. Based on our proposed CNN extension, the analysis of a spatio-temporal filtering example originated from analytical modeling of receptive fields of the visual cortex is presented and a spatio-temporal cone filter is designed and presented with numerical simulation results.

Synthesis of Nonseparable 3-D Spatiotemporal Bandpass Filters on Analog Networks

Linear cellular neural networks (CNNs) are capable of performing efficient spatiotemporal filtering operations as recursive infinite impulse response (IIR) filters. Particularly, linear CNNs can be characterized as a spatial frequency-dependent recursive temporal filter with complex coefficients. Based on a modified version of the CNN paradigm recently proposed by the authors, nonseparable spatiotemporal bandpass filters with tunable spatiotemporal passband volumes are synthesized. The filters reported here qualitatively resemble spatiotemporal receptive field models for the primary visual cortex. Numerical simulation results confirm the bandpass characteristics of our filtering network.

CELLULAR NEURAL NETWORKS FOR VIDEO COMPRESSION: AN OBJECT-ORIENTED APPROACH

Video compression technologies have recently become an integral part of the way we create, communicate and consume visual information. The aim of this Letter is to show that the Cellular Neural Network (CNN) paradigm can be exploited for obtaining accurate video compression. In particular, the Letter presents an architecture that combines CNN algorithms and H.264 codec. The compression capabilities of the devised coding system are analyzed in detail using some benchmark video sequences, and comparisons are carried out between the CNN-based approach and the H.264 codec working alone. The outcome of the analysis is that the CNN-based coding approach outperforms the H.264 codec working alone, allowing to perceive the capabilities of the CNN paradigm.

CELLULAR NEURAL NETWORKS FOR EDGE DETECTION

This Letter presents an effective edge detection technique based on the cellular neural network paradigm. The approach exploits a rigorous model of the image contours and takes into account some electrical restrictions of existing hardware implementations. The method yields accurate results, better than the ones achievable by other cellular neural network-based techniques.

New object-oriented segmentation algorithm based on the CNN paradigm

This paper illustrates a new object-oriented segmentation algorithm based on the cellular neural network (CNN) paradigm. The approach, which exploits rigorous model of the image contours, presents two remarkable features: 1) it provides more accurate segmented objects than the ones obtained by other CNN-based techniques; 2) it makes use of CNN templates that take into account the hardware characteristics imposed by the CNNUM. Results carried out for benchmark video sequences highlight the capabilities of the proposed technique.

A CNN algorithm for real time analysis of DNA microarrays

The work focuses on a new approach for DNA-chip image processing by using the cellular neural network (CNN) paradigm. In particular, the work aims to solve the main issues in automatic classification and spot validation that arise during an automatic DNA microarray analysis procedure. This contribution briefly presents the basic DNA micro-array technique and the state-of-the-art technologies, and introduces the idea of processing DNA chip data via CNNs, reporting suitable examples. The algorithm described has been developed using a high level language, dedicated to CNN universal machine (CNNUM) chip programming, and validated on a typical fluorescence image from a DNA microarray after hybridization.

Cellular neural networks: a paradigm for nonlinear spatio-temporal processing

The paradigm of Cellular Neural Networks (CNNs)is going to achieve a complete maturity. In fact, from a methodological point of view, important results on their digitally programmable analog dynamics have been gained, completed with thousands of application routines. This has encouraged the spreading of a great number of applications in the most different disciplines. Moreover, their structure, tailor made for VLSI realization, has led to the production of some chip prototypes that, embedded in a computational infrastructure, have produced the first analogic cellular computers. This completes the framework and makes it possible to realize complex spatio-temporal and filtering tasks on a time scale of microseconds. In this paper some sketches on the main aspects of CNNs, from the formal to the hardware prototype point of view, are presented together with some appealing applications to illustrate complex image, visual and spatio-temporal dynamics processing.

Object recognition in image sequences with cellular neural networks

In this paper, the application of CNN associative memories for 3D object recognition is presented. The main idea is to analyse the optical flow in an image sequence of an object. Several features of the optical flow between two succeeding images are calculated and merged to a time series of features for the whole image sequence. These features show several object specific characteristics and are used for a classification step in an object recognition system. Therefore, the feature vectors of an object set are learnt and recalled by an associative memory based on the paradigm of cellular neural networks (CNN).

Analog cellular networks for multisensor fusion and control

Analog modular architectures, derived from the computational paradigm of state-controlled cellular neural networks (SC-CNNs), are considered in this brief to process signals gathered from a distributed set of sensors. A novel design methodology for choosing the "local" system parameters so as to obtain the desired "global" signal processing function is proposed together with some theoretical results on sufficient conditions that guarantee asymptotic stability. An experimental prototype of cellular neural network for multisensor data fusion and control applications is presented and its adoption in the field of smart structures is discussed.

Analog VLSI implementation of the Help If Needed Stereopsis Algorithm

This work introduces a novel clocked analog VLSI hardware system with an optical input that performs stereopsis. An algorithm called the Help If Needed Algorithm, developed previously, is readily mapped onto an analog VLSI platform. The system fits into the cellular neural network (CNN) paradigm. The circuit components that make up the cells of the CNN are designed with the constraint that they must function effectively and fit into the space available. In order to clarify the processing pathway, the system is described at the component and system levels. Each cell has an optical input, while the output is electrical. By utilizing an optical input, an analog VLSI silicon retina first stage can be connected to the stereopsis processor completely in parallel, creating a multi-stage artificial visual system. The physical system is composed of 2.0 /spl mu/m Tinychips fabricated through MOSIS. Experimental data are presented that verify that the system performs as desired and successfully implements the Help If Needed Stereopsis Algorithm. The novel stereopsis processor is ideally suited for autonomous robots, or any application that requires a low power visual processing system.

Analogue median/average image filter based on cellularneural network paradigm

A cellular neural network-based image filter is presented, which alloys for median and mean image filtering. The circuit implements an array of 3/spl times/64 analogue processing elements (cells) and appropriate additional circuitry. Images are loaded into the circuit, are read out of the circuit serially, and are processed in real-time.

Structural analysis of stereograms for CNN depth detection

The usefulness of combining stereogram techniques with cellular neural network (CNN) analogic procedures in stereo depth extraction has been demonstrated before. Due to the local processing philosophy in the CNN paradigm, the detailed and well-established knowledge of local stereogram properties is of high importance. The paper gives an algorithmic definition for synthetic stereograms; a means for structural analysis of different types of random stereograms. In order to confine the generally propagating effect of three-dimensional (3-D) surface features into a restricted neighborhood in a stereogram, the concept of difference stereograms and the related difference surface is introduced for coding local surface variations into stereograms. The difference stereogram with its local structure, is especially suitable for CNN processing, and it is closely connected to the pattern projection technique of depth extraction.

Subband coding and image compression using CNN

The cellular neural network paradigm has found many applications in image processing. However, algorithms for image compression using CNN have scarcely been explored. CNN programmability is based on a new algorithmic style based on the spatio-temporal properties of the array. By exploiting the massive parallelism provided by CNN and the convolutional key basic instruction, a fast and efficient compression process can be achieved. This paper presents new templates and low-complexity algorithms to perform both the linear and non-linear operations needed for image compression. In this work, we have addressed all the transformation steps needed in image compression, i.e. decorrelation, bit allocation, quantization and bit extraction. From all possible compression techniques the wavelet subband coding was chosen because it is considered one of the most successful techniques for lossy compression. It allows a high compression ratio while preserving the image quality. All these advantages are implemented in the algorithms hereby presented.

Sensing the third dimension in stereo vision systems: a cellular neural networks approach

In this paper the use of the cellular neural network (CNN) paradigm is investigated for the vision-based real-time guidance of robots. This paradigm is employed in recovering information on the tridimensional structure of the environment, through the resolution of the static and the lateral motion stereo vision problems. The proposed approaches exploit the spontaneous internal energy decrease of the CNN, coding the problem in terms of an optimisation task. Results of computer simulations on some test cases for the two different issues are provided. The performance of a hardware implementation of these networks for the tasks presented is outlined.