Hopfield Neural Network Architecture Research Articles

A Context-Based Image Contrast Enhancement Using Energy Equalization With Clipping Limit.

In this paper, a new context-based image contrast enhancement process using energy curve equalization (ECE) with a clipping limit has been proposed. In a fundamental anomaly to the existing contrast enhancement practice using histogram equalization, the projected method uses the energy curve. The computation of the energy curve utilizes a modified Hopfield neural network architecture. This process embraces the image's spatial adjacency information to the energy curve. For each intensity level, the energy value is calculated and the overall energy curve appears to be smoother than the histogram. A clipping limit applies to evade the over enhancement and is chosen as the average of the mean and median value. The clipped energy curve is subdivided into three regions based on the standard deviation value. Each part of the subdivided energy curve is equalized individually, and the final enhanced image is produced by combining transfer functions computed by the equalization process. The projected scheme's qualitative and quantitative efficiency is assessed by comparing it with the conventional histogram equalization techniques with and without the clipping limit.

Exposure-Based Energy Curve Equalization for Enhancement of Contrast Distorted Images

This paper presents a novel image contrast enhancement technique that uses exposure-based energy curve equalization (ECE) with a plateau limit. In a primary deviation from the current histogram equalization process for contrast enhancement, the proposed approach uses an energy curve for the same. The energy curve is computed based on the modified Hopfield neural network architecture, which contains spatial context information. The calculated energy curve is clipped with a plateau limit computed as the average of the energy curve. The exposure threshold is computed and used to divide the clipped energy curve. The two resulting energy curves are equalized independently, and the final enhanced image is generated by integrating the images achieved by transforming the equalized energy curves. The performance of the proposed method is evaluated on a variety of low contrast images. The subjective and objective evaluations of the proposed method are compared with the various histogram equalization (HE) based methods and other state-of-the-art methods to exemplify the effectiveness.

Spatio-Temporal Data Fusion for Satellite Images Using Hopfield Neural Network

Spatio-temporal data fusion refers to the technique of combining high temporal resolution from coarse satellite images and high spatial resolution from fine satellite images. However, data availability remains a major limitation in algorithm development. Existing spatio-temporal data fusion algorithms require at least one known image pair between the fine and coarse resolution image. However, data which come from two different satellite platforms do not necessarily have an overlap in their overpass times, hence restricting the application of spatio-temporal data fusion. In this paper, a new algorithm named Hopfield Neural Network SPatio-tempOral daTa fusion model (HNN-SPOT) is developed by utilizing the optimization concept in the Hopfield neural network (HNN) for spatio-temporal image fusion. The algorithm derives a synthesized fine resolution image from a coarse spatial resolution satellite image (similar to downscaling), with the use of one fine resolution image taken on an arbitrary date and one coarse image taken on a predicted date. The HNN-SPOT particularly addresses the problem when the fine resolution and coarse resolution images are acquired from different satellite overpass times over the same geographic extent. Both simulated datasets and real datasets over Hong Kong and Australia have been used in the evaluation of HNN-SPOT. Results showed that HNN-SPOT was comparable with an existing fusion algorithm, the spatial and temporal adaptive reflectance fusion model (STARFM). HNN-SPOT assumes consistent spatial structure for the target area between the date of data acquisition and the prediction date. Therefore, it is more applicable to geographical areas with little or no land cover change. It is shown that HNN-SPOT can produce accurate fusion results with >90% of correlation coefficient over consistent land covers. For areas that have undergone land cover changes, HNN-SPOT can still produce a prediction about the outlines and the tone of the features, if they are large enough to be recorded in the coarse resolution image at the prediction date. HNN-SPOT provides a relatively new approach in spatio-temporal data fusion, and further improvements can be made by modifying or adding new goals and constraints in its HNN architecture. Owing to its lower demand for data prerequisites, HNN-SPOT is expected to increase the applicability of fine-scale applications in remote sensing, such as environmental modeling and monitoring.

The stability analysis for a novel feedback neural network with partial connection

This paper develops a new Partially Feedback Neural Network with partial connection, which is so-called “Partially Connected Feedback Neural Network” (PCFNN). The information capacity improves and there is more hidden information for partially connected systems because the connections between neurons are random and there can be more than one layer. The proving of the convergence of PCFNN is provided. Owing to the complexities in partially connected systems, two theorems of its stability are proved theoretically by constructing a novel energy function expectation. Three examples are provided to simulate various conditions of stability by constructing different activation functions and weight matrixes. The simulation results show that this novel neural network is stable under different conditions. The expressive space of the network architecture is also much larger than the original Hopfield neural network architecture in the partially connected neural network architecture.

A Context-Sensitive Technique for Unsupervised Change Detection Based on Hopfield-Type Neural Networks

In this paper, we propose a context-sensitive technique for unsupervised change detection in multitemporal remote sensing images. This technique is based on a modified Hopfield neural network architecture designed to model spatial correlation between neighboring pixels of the difference image produced by comparing images acquired on the same area at different times. Each spatial position in the considered scene is represented by a neuron in the Hopfield network that is connected only to its neighboring units. These connections model the spatial correlation between neighboring pixels and are associated with a context-sensitive energy function that represents the overall status of the network. Change detection maps are obtained by iteratively updating the output status of the neurons until a minimum of the energy function is reached and the network assumes a stable state. A simple heuristic thresholding procedure is presented and adopted for initializing the network. The proposed change detection technique is unsupervised and distribution free. Experimental results carried out on two multispectral and multitemporal remote sensing images confirm the effectiveness of the proposed technique

Image feature extraction by dynamic neural filtering and phase-only joint transform correlation

Neural network-based image processing algorithms present numerous advantages due to their supervised adjustable properties. Among various neural network architectures, dynamic neural networks, Hopfield and Cellular networks, have been found inherently suitable for filtering applications. Combining supervised and filtering features of dynamic neural networks, this paper presents dynamic neural filtering technique based on Hopfield neural network architecture. The filtering technique has also been implemented by using phase-only joint transform correlation (POJTC) for optical image processing applications. Filtering structure is basically similar to the Hopfield neural network structure except for the adjustable filter mask and 2D convolution operation instead of weight matrix operations. The dynamic neural filtering architecture has learnable properties by back-propagation learning algorithm. POJTC presents significant advantages to achieve the operation of summing the cross-correlation of bipolar data by phase-encoding bipolar data in parallel. The image feature extraction performance of the proposed optical system is reported for various image processing applications using a simulation program.

A binary Hopfield neural network with hysteresis for large crossbar packet-switches

In this paper, we propose a hysteretic Hopfield neural network architecture for efficiently solving crossbar switch problems. A binary Hopfield neural network architecture with hysteresis binary neurons and its collective computational properties are studied. The network architecture is applied to a crossbar switch problem and results of computer simulations are presented and used to illustrate the computation power of the network architecture.

A positively self-feedbacked Hopfield neural network architecture for crossbar switching

We propose a positively self-feedbacked Hopfield neural network architecture for efficiently solving crossbar switch problem. A binary Hopfield neural network architecture with additional positive self-feedbacks and its collective computational properties are studied. It is proved theoretically and confirmed by simulating the randomly generated Hopfield neural network with positive self-feedbacks that the emergent collective properties of the original Hopfield neural network also are present in this network architecture. The network architecture is applied to crossbar switching and results of computer simulations are presented and used to illustrate the computation power of the network architecture. The simulation results show that the Hopfield neural network architecture with positive self-feedbacks is much better than the previous works including the original Hopfield neural network architecture, Troudet's architecture and maximum neural network for crossbar switching in terms of both the computation time and the solution quality.

Systolic implementation of 2D block-based Hopfield neural network for efficient pattern association

A systolic array implementation of block-based Hopfield neural network architecture using completely digital circuits is presented in this paper. The design is based on modelling the energy equation of Hopfield neural network to a systolic (or modular) form. It is shown mathematically that the modified energy equation converges in all circumstances. In addition, it is shown that the architecture provides massive parallelism and can be extended to a larger network by cascading identical chips. The performance of the proposed architecture is evaluated by applying various binary inputs and it is observed that it exhibits the same characteristics of block-based Hopfield neural network in terms of convergence and pattern association.

A restoration algorithm for P-32 and Y-90 bremsstrahlung emission nuclear imaging: a wavelet-neural network approach.

A novel wavelet-based neural network (WNN) filter is proposed for image restoration as required for imaging of beta emitters by bremsstrahlung detection using a gamma camera. Quantitative imaging of beta emitters is important for the in vivo management of antibody therapy using either P-32 or Y-90. The theoretical basis for the general case for M-channel multiresolution wavelet decomposition of the nuclear image into different subimages is developed with the objective of isolating the signal from noise. A modified Hopfield neural network (NN) architecture is then used for multichannel image restoration using the dominant signal subimages. The NN model avoids the common inverse problem associated with other image restoration filters such as the Wiener filter. The relative performance of the WNN for image restoration, for M = 2 channel, is compared to a previously reported order statistic neural network hybrid (OSNNH) filter. Initially simulated degraded images of known structures with different noise levels are used. Quantitative metrics such as the normalized mean square error (NMSE) and signal-to-noise ratio (SNR) are used to compare filter performance. The WNN yields comparable results for image restoration with suggested slightly better performance for the images with higher noise levels as often encountered in bremsstrahlung detection. Attenuation measurements were performed using two radionuclides, 32P and 90Y as required for calibration of the gamma camera for quantitative measurements. Similar values for an effective attenuation coefficient were observed for the restored images using the OSNNH filters (32P: mu = 0.122 cm-1, 90Y: mu = 0.135 cm-1) and WNN (32P: mu = 0.122 cm-1, 90Y: mu = 0.135 cm-1) filters with slightly higher values obtained for the raw data (32P: mu = 0.142 cm-1, 90Y: mu = 0.142 cm-1) for a 3.5-cm source size. The WNN, however, was computationally more efficient by a factor of 4 to 6 compared to the OSNNH filter. The filter architecture, in turn, is also optimum for parallel processing or VLSI implementation as required for planar and particularly for SPECT mode of detection.

Wavelet-based neural network with fuzzy-logic adaptivity for nuclear image restoration

A novel wavelet-based neural network with fuzzy-logic adaptivity (WNNFA) is proposed for image restoration using a nuclear medicine gamma camera based on the measured system point spread function. The objective is to restore image degradation due to photon scattering and collimator photon penetration with the gamma camera and allow improved quantitative external measurements of radionuclides in vivo. The specific clinical model proposed is the imaging of bremsstrahlung radiation using /sup 32/P and /sup 90/Y. The theoretical basis for four-channel multiresolution wavelet decomposition of the nuclear image into different subimages is developed with the objective of isolating the signal from noise. A fuzzy rule is generated to train a membership function using least mean squares to obtain an optimal balance between image restoration and the stability of the neural network, while maintaining a linear response for the camera to radioactivity dose. A multichannel modified Hopfield neural network architecture is then proposed for multichannel image restoration using the dominant signal subimages.

Use of Hopfield neural networks in optimal guidance

A Hopfield neural network architecture is developed to solve the optimal control problem for homing missile guidance. A linear quadratic optimal control problem is formulated in the form of an efficient parallel computing device known as a Hopfield neural network. Convergence of the Hopfield network is analyzed from a theoretical perspective, showing that the network, as a dynamical system approaches a unique fixed point which is the solution to the optimal control problem at any instant during the missile pursuit. Several target-intercept scenarios are provided to demonstrate the use of the recurrent feedback neural net formulation.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Neural network architecture for crossbar switch control

A Hopfield neural network architecture for the real-time control of a crossbar switch for switching pockets at maximum throughput is proposed. The network performance and processing time are derived from a numerical simulation of the transitions of the neural network. A method is proposed to optimize electronic component parameters and synaptic connections, and it is fully illustrated by the computer simulation of a VLSI implementation of 4*4 neural net controller. The extension to larger size crossbars is demonstrated through the simulation of an 8*8 crossbar switch controller, where the performance of the neural computation is discussed in relation to electronic noise and inhomogeneities of network components. >