Bayesian optimized multimodal deep hybrid learning approach for tomato leaf disease classification

Manual identification of tomato leaf diseases is a time-consuming and laborious process that may lead to inaccurate results without professional assistance. Therefore, an automated, early, and precise leaf disease recognition system is essential for farmers to ensure the quality and quantity of tomato production by providing timely interventions to mitigate disease spread. In this study, we have proposed seven robust Bayesian optimized deep hybrid learning models leveraging the synergy between deep learning and machine learning for the automated classification of ten types of tomato leaves (nine diseased and one healthy). We customized the popular Convolutional Neural Network (CNN) algorithm for automatic feature extraction due to its ability to capture spatial hierarchies of features directly from raw data and classical machine learning techniques [Random Forest (RF), XGBoost, GaussianNB (GNB), Support Vector Machines (SVM), Multinomial Logistic Regression (MLR), K-Nearest Neighbor (KNN)], and stacking for classifications. Additionally, the study incorported a Boruta feature filtering layer to capture the statistically significant features. The standard, research-oriented PlantVillage dataset was used for the performance testing, which facilitates benchmarking against prior research and enables meaningful comparisons of classification performance across different approaches. We utilized a variety of statistical classification metrics to demonstrate the robustness of our models. Using the CNN-Stacking model, this study achieved the highest classification performance among the seven hybrid models. On an unseen dataset, this model achieved average precision, recall, f1-score, mcc, and accuracy values of 98.527%, 98.533%, 98.527%, 98.525%, and 98.268%, respectively. Our study requires only 0.174 s of testing time to correctly identify noisy, blurry, and transformed images. This indicates our approach's time efficiency and generalizability in images captured under challenging lighting conditions and with complex backgrounds. Based on the comparative analysis, our approach is superior and computationally inexpensive compared to the existing studies. This work will aid in developing a smartphone app to offer farmers a real-time disease diagnosis tool and management strategies.

Underwater Image Enhancement Based on Luminance Reconstruction by Multi-Resolution Fusion of RGB Channels

Underwater image enhancement technology is crucial for the human exploration and exploitation of marine resources. The visibility of underwater images is affected by visible light attenuation. This paper proposes an image reconstruction method based on the decomposition-fusion of multi-channel luminance data to enhance the visibility of underwater images. The proposed method is a single-image approach to cope with the condition that underwater paired images are difficult to obtain. The original image is first divided into its three RGB channels. To reduce artifacts and inconsistencies in the fused images, a multi-resolution fusion process based on the Laplace-Gaussian pyramid guided by a weight map is employed. Image saliency analysis and mask sharpening methods are also introduced to color-correct the fused images. The results indicate that the method presented in this paper effectively enhances the visibility of dark regions in the original image and globally improves its color, contrast, and sharpness compared to current state-of-the-art methods. Our method can enhance underwater images in engineering practice, laying the foundation for in-depth research on underwater images.

Deep learning based on small sample dataset: prediction of dielectric properties of SrTiO3-type perovskite with doping modification

The perovskite crystal structure represents a semiconductor material poised for widespread application, underpinned by attributes encompassing heightened efficiency, cost-effectiveness and remarkable flexibility. Notably, strontium titanate (SrTiO3)-type perovskite, a prototypical ferroelectric dielectric material, has emerged as a pre-eminent matrix material for enhancing the energy storage capacity of perovskite. Typically, the strategy involves augmenting its dielectric constant through doping to enhance energy storage density. However, SrTiO3 doping data are plagued by significant dispersion, and the small sample size poses a formidable research hurdle, hindering the investigation of dielectric property and energy storage density enhancements. This study endeavours to address this challenge, our foundation lies in the compilation of 200 experimental records related to SrTiO3-type perovskite doping, constituting a small dataset. Subsequently, an interactive framework harnesses deep neural network models and a one-dimensional convolutional neural network model to predict and scrutinize the dataset. Distinctively, the mole percentage of doping elements exclusively serves as input features, yielding significantly enhanced accuracy in dielectric performance prediction. Lastly, rigorous comparisons with traditional machine learning models, specifically gradient boosting regression, validate the superiority and reliability of deep learning models. This research advances a novel, effective methodology and offers a valuable reference for designing and optimizing perovskite energy storage materials.

EfficientRMT-Net-An Efficient ResNet-50 and Vision Transformers Approach for Classifying Potato Plant Leaf Diseases

The primary objective of this study is to develop an advanced, automated system for the early detection and classification of leaf diseases in potato plants, which are among the most cultivated vegetable crops worldwide. These diseases, notably early and late blight caused by Alternaria solani and Phytophthora infestans, significantly impact the quantity and quality of global potato production. We hypothesize that the integration of Vision Transformer (ViT) and ResNet-50 architectures in a new model, named EfficientRMT-Net, can effectively and accurately identify various potato leaf diseases. This approach aims to overcome the limitations of traditional methods, which are often labor-intensive, time-consuming, and prone to inaccuracies due to the unpredictability of disease presentation. EfficientRMT-Net leverages the CNN model for distinct feature extraction and employs depth-wise convolution (DWC) to reduce computational demands. A stage block structure is also incorporated to improve scalability and sensitive area detection, enhancing transferability across different datasets. The classification tasks are performed using a global average pooling layer and a fully connected layer. The model was trained, validated, and tested on custom datasets specifically curated for potato leaf disease detection. EfficientRMT-Net's performance was compared with other deep learning and transfer learning techniques to establish its efficacy. Preliminary results show that EfficientRMT-Net achieves an accuracy of 97.65% on a general image dataset and 99.12% on a specialized Potato leaf image dataset, outperforming existing methods. The model demonstrates a high level of proficiency in correctly classifying and identifying potato leaf diseases, even in cases of distorted samples. The EfficientRMT-Net model provides an efficient and accurate solution for classifying potato plant leaf diseases, potentially enabling farmers to enhance crop yield while optimizing resource utilization. This study confirms our hypothesis, showcasing the effectiveness of combining ViT and ResNet-50 architectures in addressing complex agricultural challenges.

Digital image processing realized by memristor-based technologies

Today performance and operational efficiency of computer systems on digital image processing are exacerbated owing to the increased complexity of image processing. It is also difficult for image processors based on complementary metal-oxide-semiconductor (CMOS) transistors to continuously increase the integration density, causing by their underlying physical restriction and economic costs. However, such obstacles can be eliminated by non-volatile resistive memory technologies (known as memristors), arising from their compacted area, speed, power consumption high efficiency, and in-memory computing capability. This review begins with presenting the image processing methods based on pure algorithm and conventional CMOS-based digital image processing strategies. Subsequently, current issues faced by digital image processing and the strategies adopted for overcoming these issues, are discussed. The state-of-the-art memristor technologies and their challenges in digital image processing applications are also introduced, such as memristor-based image compression, memristor-based edge and line detections, and voice and image recognition using memristors. This review finally envisages the prospects for successful implementation of memristor devices in digital image processing.

An efficient registration-based approach for retinal blood vessel segmentation using generalized Pareto and fatigue pdf

Segmentation of Retinal Blood Vessel (RBV) extraction in the retina images and Registration of segmented RBV structure is implemented to identify changes in vessel structure by ophthalmologists in diagnosis of various illnesses like Glaucoma, Diabetes, and Hypertension's. The Retinal Blood Vessel provides blood to the inner retinal neurons, RBV are located mainly in internal retina but it may partly in the ganglion cell layer, following network failure haven't been identified with past methods. Classifications of accurate RBV and Registration of segmented blood vessels are challenging tasks in the low intensity background of Retinal Image. So, we projected a novel approach of segmentation of RBV extraction used matched filter of Generalized Pareto Probability Distribution Function (pdf) and Registration approach on feature-based segmented retinal blood vessel of Binary Robust Invariant Scalable Key point (BRISK). The BRISK provides the predefined sampling pattern as compared to Pdf. The BRISK feature is implemented for attention point recognition & matching approach for change in vessel structure. The proposed approaches contain 3 levels: pre-processing, matched filter-based Generalized Pareto pdf as a source along with the novel approach of fatigue pdf as a target, and BRISK framework is used for Registration on segmented retinal images of supply & intention images. This implemented system's performance is estimated in experimental analysis by the Average accuracy, Normalized Cross-Correlation (NCC), and computation time process of the segmented retinal source and target image. The NCC is main element to give more statistical information about retinal image segmentation. The proposed approach of Generalized Pareto value pdf has Average Accuracy of 95.21%, NCC of both image pairs is 93%, and Average accuracy of Registration of segmented source images and the target image is 98.51% respectively. The proposed approach of average computational time taken is around 1.4 s, which has been identified on boundary condition of Pdf function.

A robust deep learning approach for tomato plant leaf disease localization and classification

Tomato plants' disease detection and classification at the earliest stage can save the farmers from expensive crop sprays and can assist in increasing the food quantity. Although, extensive work has been presented by the researcher for the tomato plant disease classification, however, the timely localization and identification of various tomato leaf diseases is a complex job as a consequence of the huge similarity among the healthy and affected portion of plant leaves. Furthermore, the low contrast information between the background and foreground of the suspected sample has further complicated the plant leaf disease detection process. To deal with the aforementioned challenges, we have presented a robust deep learning (DL)-based approach namely ResNet-34-based Faster-RCNN for tomato plant leaf disease classification. The proposed method includes three basic steps. Firstly, we generate the annotations of the suspected images to specify the region of interest (RoI). In the next step, we have introduced ResNet-34 along with Convolutional Block Attention Module (CBAM) as a feature extractor module of Faster-RCNN to extract the deep key points. Finally, the calculated features are utilized for the Faster-RCNN model training to locate and categorize the numerous tomato plant leaf anomalies. We tested the presented work on an accessible standard database, the PlantVillage Kaggle dataset. More specifically, we have obtained the mAP and accuracy values of 0.981, and 99.97% respectively along with the test time of 0.23 s. Both qualitative and quantitative results confirm that the presented solution is robust to the detection of plant leaf disease and can replace the manual systems. Moreover, the proposed method shows a low-cost solution to tomato leaf disease classification which is robust to several image transformations like the variations in the size, color, and orientation of the leaf diseased portion. Furthermore, the framework can locate the affected plant leaves under the occurrence of blurring, noise, chrominance, and brightness variations. We have confirmed through the reported results that our approach is robust to several tomato leaf diseases classification under the varying image capturing conditions. In the future, we plan to extend our approach to apply it to other parts of plants as well.

DeeProPre: A promoter predictor based on deep learning

The promoter is a DNA sequence recognized, bound and transcribed by RNA polymerase. It is usually located at the upstream or 5'end of the transcription start site (TSS). Studies have shown that the structure of the promoter affects its affinity for RNA polymerase, thus affecting the level of gene expression. Therefore, the correct identification of core promoter and common structural gene is of great significance in the field of biomedicine. At present, many methods have been proposed to improve the accuracy of promoter recognition, but the performances still need to be further improved. In this study, a deep learning algorithm (DeeProPre) based on bidirectional long short-term memory (BiLSTM) and convolutional neural network (CNN) was proposed. Firstly, the supervised embedding layer was applied to map the sequence to a high-dimensional space. Secondly, two 1D convolutional layers, BiLSTM and attentional mechanism layer were used for extracting features. Finally, the full connection layer activated by Sigmoid function was used to obtain the probability of classification into target categories. This model can identify the promoter region of eukaryotes with high accuracy, providing an analytical basis for further understanding of promoter physiological functions and studies of gene transcription mechanisms. The source code of DeeProPre is freely available at https://github.com/zzwwmmm/DeeProPre/tree/master.

Application of GIS Technology-Supported Cross Media Fusion Method Based on Deep Learning in Landscape Performance Evaluation

GIS technology can provide reasonable and sustainable data support for landscape planning and ecological development and make wetland landscape planning consider the spatial layout of landscape and the optimal allocation of resources more. The key technologies of cross media intelligence mainly focus on intelligent information retrieval, analysis and reasoning, knowledge map construction, and intelligent storage. Convolutional neural network (CNN), as one of the representative algorithms of deep learning, plays an important role in retrieving landscape data and extracting image and text features across media. Further retrieval of media data, in-depth text processing, and image feature data extraction are realized by using deep learning technology, and comprehensive in-depth analysis is carried out by combining landscape plane images, three-dimensional images, and vector information in GIS technology. Provide quantitative information for the evaluation system of human landscape, economy, history, and region, so as to formulate a scientific and reasonable performance evaluation system.

DCNet: DenseNet-77-based CornerNet model for the tomato plant leaf disease detection and classification

Early recognition of tomato plant leaf diseases is mandatory to improve the food yield and save agriculturalists from costly spray procedures. The correct and timely identification of several tomato plant leaf diseases is a complicated task as the healthy and affected areas of plant leaves are highly similar. Moreover, the incidence of light variation, color, and brightness changes, and the occurrence of blurring and noise on the images further increase the complexity of the detection process. In this article, we have presented a robust approach for tackling the existing issues of tomato plant leaf disease detection and classification by using deep learning. We have proposed a novel approach, namely the DenseNet-77-based CornerNet model, for the localization and classification of the tomato plant leaf abnormalities. Specifically, we have used the DenseNet-77 as the backbone network of the CornerNet. This assists in the computing of the more nominative set of image features from the suspected samples that are later categorized into 10 classes by the one-stage detector of the CornerNet model. We have evaluated the proposed solution on a standard dataset, named PlantVillage, which is challenging in nature as it contains samples with immense brightness alterations, color variations, and leaf images with different dimensions and shapes. We have attained an average accuracy of 99.98% over the employed dataset. We have conducted several experiments to assure the effectiveness of our approach for the timely recognition of the tomato plant leaf diseases that can assist the agriculturalist to replace the manual systems.

Custom CornerNet: a drone-based improved deep learning technique for large-scale multiclass pest localization and classification

Insect pests are among the most critical factors affecting crops and result in a severe reduction in food yield. At the same time, early and accurate identification of insect pests can assist farmers in taking timely preventative steps to reduce financial losses and improve food quality. However, the manual inspection process is a daunting and time-consuming task due to visual similarity between various insect species. Moreover, sometimes it is difficult to find an experienced professional for the consultation. To deal with the problems of manual inspection, we have presented an automated framework for the identification and categorization of insect pests using deep learning. We proposed a lightweight drone-based approach, namely a custom CornerNet approach with DenseNet-100 as a base network. The introduced framework comprises three phases. The region of interest is initially acquired by developing sample annotations later used for model training. A custom CornerNet is proposed in the next phase by employing the DenseNet-100 for deep keypoints computation. The one-stage detector CornerNet identifies and categorizes several insect pests in the final step. The DenseNet network improves the capacity of feature representation by connecting the feature maps from all of its preceding layers and assists the CornerNet model in detecting insect pests as paired vital points. We assessed the performance of the proposed model on the standard IP102 benchmark dataset for pest recognition which is challenging in terms of pest size, color, orientation, category, chrominance, and lighting variations. Both qualitative and quantitative experimental results showed the effectiveness of our approach for identifying target insects in the field with improved accuracy and recall rates.

Research on Blended Teaching of Flipped Classroom Based on CNN-SSA-Bi-LSTM Deep Learning Model Computer Media

Aiming at the problem that the influencing factors of computer media flipped classroom hybrid teaching lead to the teaching effect not reaching the expected, this study proposes an ultra-short-term prediction model based on CNN-SSA-Bi-LSTM. CNN-SSA-Bi-LSTM is used to flip the study of mixed teaching in the classroom. This method constructs a one-dimensional convolutional neural network, performs data fusion and feature transformation on multiple key variables, and then constructs a two-way long-term short-term memory network prediction model, which realizes a 45-minute classroom for ultra-short-term prediction of the future. In addition, data optimization is performed through SSA to improve the predictive effect of the CNN-Bi-LSTM model. Experimental results show that compared with the traditional machine learning method, the proposed prediction model can effectively improve the prediction accuracy of the ultra-short-term classroom effect, and the relative variance of the continuous model is increased by 16.22%. High prediction accuracy and low error prove that CNN-SSA-Bi-LSTM deep learning model has strong application prospects in the research of flipped classroom hybrid teaching.

Wave Loss: A Topographic Metric for Image Segmentation

The solution of segmentation problems with deep neural networks requires a well-defined loss function for comparison and network training. In most network training approaches, only area-based differences that are of differing pixel matter are considered; the distribution is not.Our brain can compare complex objects with ease and considers both pixel level and topological differences simultaneously and comparison between objects requires a properly defined metric that determines similarity between them considering changes both in shape and values. In past years, topographic aspects were incorporated in loss functions where either boundary pixels or the ratio of the areas were employed in difference calculation. In this paper we will show how the application of a topographic metric, called wave loss, can be applied in neural network training and increase the accuracy of traditional segmentation algorithms. Our method has increased segmentation accuracy by 3% on both the Cityscapes and Ms-Coco datasets, using various network architectures.

DNA-binding protein prediction based on deep transfer learning

The study of DNA binding proteins (DBPs) is of great importance in the biomedical field and plays a key role in this field. At present, many researchers are working on the prediction and detection of DBPs. Traditional DBP prediction mainly uses machine learning methods. Although these methods can obtain relatively high pre-diction accuracy, they consume large quantities of human effort and material resources. Transfer learning has certain advantages in dealing with such prediction problems. Therefore, in the present study, two features were extracted from a protein sequence, a transfer learning method was used, and two classical transfer learning algorithms were compared to transfer samples and construct data sets. In the final step, DBPs are detected by building a deep learning neural network model in a way that uses attention mechanisms.

A Dense Feature Pyramid Network for Remote Sensing Object Detection

In recent years, object detection in remote sensing images has become a popular topic in computer vision research. However, there are various problems in remote sensing object detection, such as complex scenes, small objects in large fields of view, and multi-scale object in different categories. To address these issues, we propose DFPN-YOLO, a dense feature pyramid network for remote sensing object detection. To address difficulties in detecting small objects in large scenes, we add a larger detection layer on top of the three detection layers of YOLOv3, and we propose Dense-FPN, a dense feature pyramid network structure that enables all four detection layers to combine semantic information before sampling and after sampling to improve the performance of object detection at different scales. In addition, we add an attention module in the residual blocks of the backbone to allow the network to quickly extract key feature information in complex scenes. The results show that the mean average precision (mAP) of our method on the RSOD datasets reached 92%, which is 8% higher than the mAP of YOLOv3, and the mAP increased from 62.41% on YOLOv3 to 69.33% with our method on the DIOR datasets, outperforming even YOLOv4.

A novel deep learning method for detection and classification of plant diseases

The agricultural production rate plays a pivotal role in the economic development of a country. However, plant diseases are the most significant impediment to the production and quality of food. The identification of plant diseases at an early stage is crucial for global health and wellbeing. The traditional diagnosis process involves visual assessment of an individual plant by a pathologist through on-site visits. However, manual examination for crop diseases is restricted because of less accuracy and the small accessibility of human resources. To tackle such issues, there is a demand to design automated approaches capable of efficiently detecting and categorizing numerous plant diseases. Precise identification and classification of plant diseases is a tedious job due because of the occurrence of low-intensity information in the image background and foreground, the huge color resemblance in the healthy and diseased plant areas, the occurrence of noise in the samples, and changes in the position, chrominance, structure, and size of plant leaves. To tackle the above-mentioned problems, we have introduced a robust plant disease classification system by introducing a Custom CenterNet framework with DenseNet-77 as a base network. The presented method follows three steps. In the first step, annotations are developed to get the region of interest. Secondly, an improved CenterNet is introduced in which DenseNet-77 is proposed for deep keypoints extraction. Finally, the one-stage detector CenterNet is used to detect and categorize several plant diseases. To conduct the performance analysis, we have used the PlantVillage Kaggle database, which is the standard dataset for plant diseases and challenges in terms of intensity variations, color changes, and differences found in the shapes and sizes of leaves. Both the qualitative and quantitative analysis confirms that the presented method is more proficient and reliable to identify and classify plant diseases than other latest approaches.

System-Theoretic Methods for Designing Bio-Inspired Mem-Computing Memristor Cellular Nonlinear Networks

The introduction of nano-memristors in electronics may allow to boost the performance of integrated circuits beyond the Moore era, especially in view of their extraordinary capability to process and store data in the very same physical volume. However, recurring to nonlinear system theory is absolutely necessary for the development of a systematic approach to memristive circuit design. In fact, the application of linear system-theoretic techniques is not suitable to explore thoroughly the rich dynamics of resistance switching memories, and designing circuits without a comprehensive picture of the nonlinear behaviour of these devices may lead to the realization of technical systems failing to operate as desired. Converting traditional circuits to memristive equivalents may require the adaptation of classical methods from nonlinear system theory. This paper extends the theory of time- and space-invariant standard cellular nonlinear networks with first-order processing elements for the case where a single non-volatile memristor is inserted in parallel to the capacitor in each cell. A novel nonlinear system-theoretic method allows to draw a comprehensive picture of the dynamical phenomena emerging in the memristive mem-computing array, beautifully illustrated in the so-called Primary Mosaic for the class of uncoupled memristor cellular nonlinear networks. Employing this new analysis tool it is possible to elucidate, with the support of illustrative examples, how to design variability-tolerant bio-inspired cellular nonlinear networks with second-order memristive cells for the execution of computing tasks or of memory operations. The capability of the class of memristor cellular nonlinear networks under focus to store and process information locally, without the need to insert additional memory units in each cell, may allow to increase considerably the spatial resolution of state-of-the-art purely CMOS sensor-processor arrays. This is of great appeal for edge computing applications, especially since the Internet-of-Things industry is currently calling for the realization of miniaturized, lightweight, low-power, and high-speed mem-computers with sensing capability on board.

A Novel Recurrent Neural Network-Based Ultra-Fast, Robust, and Scalable Solver for Inverting a "Time-Varying Matrix"

The concept presented in this paper is based on previous dynamical methods to realize a time-varying matrix inversion. It is essentially a set of coupled ordinary differential equations (ODEs) which does indeed constitute a recurrent neural network (RNN) model. The coupled ODEs constitute a universal modeling framework for realizing a matrix inversion provided the matrix is invertible. The proposed model does converge to the inverted matrix if the matrix is invertible, otherwise it converges to an approximated inverse. Although various methods exist to solve a matrix inversion in various areas of science and engineering, most of them do assume that either the time-varying matrix inversion is free of noise or they involve a denoising module before starting the matrix inversion computation. However, in the practice, the noise presence issue is a very serious problem. Also, the denoising process is computationally expensive and can lead to a violation of the real-time property of the system. Hence, the search for a new 'matrix inversion' solving method inherently integrating noise-cancelling is highly demanded. In this paper, a new combined/extended method for time-varying matrix inversion is proposed and investigated. The proposed method is extending both the gradient neural network (GNN) and the Zhang neural network (ZNN) concepts. Our new model has proven that it has exponential stability according to Lyapunov theory. Furthermore, when compared to the other previous related methods (namely GNN, ZNN, Chen neural network, and integration-enhanced Zhang neural network or IEZNN) it has a much better theoretical convergence speed. To finish, all named models (the new one versus the old ones) are compared through practical examples and both their respective convergence and error rates are measured. It is shown/observed that the novel/proposed method has a better practical convergence rate when compared to the other models. Regarding the amount of noise, it is proven that there is a very good approximation of the matrix inverse even in the presence of noise.

Switch Elements with S-Shaped Current-Voltage Characteristic in Models of Neural Oscillators

In this paper, we present circuit solutions based on a switch element with the S-type I–V characteristic implemented using the classic FitzHugh–Nagumo and FitzHugh–Rinzel models. Using the proposed simplified electrical circuits allows the modeling of the integrate-and-fire neuron and burst oscillation modes with the emulation of the mammalian cold receptor patterns. The circuits were studied using the experimental I–V characteristic of an NbO2 switch with a stable section of negative differential resistance (NDR) and a VO2 switch with an unstable NDR, considering the temperature dependences of the threshold characteristics. The results are relevant for modern neuroelectronics and have practical significance for the introduction of the neurodynamic models in circuit design and the brain–machine interface. The proposed systems of differential equations with the piecewise linear approximation of the S-type I–V characteristic may be of scientific interest for further analytical and numerical research and development of neural networks with artificial intelligence.

A Method for Evaluating Chimeric Synchronization of Coupled Oscillators and Its Application for Creating a Neural Network Information Converter

This paper presents a new method for evaluating the synchronization of quasi-periodic oscillations of two oscillators, termed “chimeric synchronization”. The family of metrics is proposed to create a neural network information converter based on a network of pulsed oscillators. In addition to transforming input information from digital to analogue, the converter can perform information processing after training the network by selecting control parameters. In the proposed neural network scheme, the data arrives at the input layer in the form of current levels of the oscillators and is converted into a set of non-repeating states of the chimeric synchronization of the output oscillator. By modelling a thermally coupled VO2-oscillator circuit, the network setup is demonstrated through the selection of coupling strength, power supply levels, and the synchronization efficiency parameter. The distribution of solutions depending on the operating mode of the oscillators, sub-threshold mode, or generation mode are revealed. Technological approaches for the implementation of a neural network information converter are proposed, and examples of its application for image filtering are demonstrated. The proposed method helps to significantly expand the capabilities of neuromorphic and logical devices based on synchronization effects.

Retinal blood vessels segmentation by using Gumbel probability distribution function based matched filter

BACKGROUND AND OBJECTIVE

Retinal blood vessel segmentation is a prominent task for the diagnosis of various retinal pathology such as hypertension, diabetes, glaucoma, etc. In this paper, a novel matched filter approach with the Gumbel probability distribution function as its kernel is introduced to improve the performance of retinal blood vessel segmentation.

METHODS

Before applying the proposed matched filter, the input retinal images are pre-processed. During pre-processing stage principal component analysis (PCA) based gray scale conversion followed by contrast limited adaptive histogram equalization (CLAHE) are applied for better enhancement of retinal image. After that an exhaustive experiments have been conducted for selecting the appropriate value of parameters to design a new matched filter. The post-processing steps after applying the proposed matched filter include the entropy based optimal thresholding and length filtering to obtain the segmented image.

RESULTS

For evaluating the performance of proposed approach, the quantitative performance measures, an average accuracy, average true positive rate (ATPR), and average false positive rate (AFPR) are calculated. The respective values of the quantitative performance measures are 0.9522, 0.7594, 0.0292 for DRIVE data set and 0.9270, 0.7939, 0.0624 for STARE data set. To justify the effectiveness of proposed approach, receiver operating characteristic (ROC) curve is plotted and the average area under the curve (AUC) is calculated. The average AUC for DRIVE and STARE data sets are 0.9287 and 0.9140 respectively.

CONCLUSIONS

The obtained experimental results confirm that the proposed approach performance better with respect to other prominent Gaussian distribution function and Cauchy PDF based matched filter approaches.

Asymmetric continuous-time neural networks without local traps for solving constraint satisfaction problems

There has been a long history of using neural networks for combinatorial optimization and constraint satisfaction problems. Symmetric Hopfield networks and similar approaches use steepest descent dynamics, and they always converge to the closest local minimum of the energy landscape. For finding global minima additional parameter-sensitive techniques are used, such as classical simulated annealing or the so-called chaotic simulated annealing, which induces chaotic dynamics by addition of extra terms to the energy landscape. Here we show that asymmetric continuous-time neural networks can solve constraint satisfaction problems without getting trapped in non-solution attractors. We concentrate on a model solving Boolean satisfiability (k-SAT), which is a quintessential NP-complete problem. There is a one-to-one correspondence between the stable fixed points of the neural network and the k-SAT solutions and we present numerical evidence that limit cycles may also be avoided by appropriately choosing the parameters of the model. This optimal parameter region is fairly independent of the size and hardness of instances, this way parameters can be chosen independently of the properties of problems and no tuning is required during the dynamical process. The model is similar to cellular neural networks already used in CNN computers. On an analog device solving a SAT problem would take a single operation: the connection weights are determined by the k-SAT instance and starting from any initial condition the system searches until finding a solution. In this new approach transient chaotic behavior appears as a natural consequence of optimization hardness and not as an externally induced effect.

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.

Design of silicon brains in the nano-CMOS era: spiking neurons, learning synapses and neural architecture optimization

We present a design framework for neuromorphic architectures in the nano-CMOS era. Our approach to the design of spiking neurons and STDP learning circuits relies on parallel computational structures where neurons are abstracted as digital arithmetic logic units and communication processors. Using this approach, we have developed arrays of silicon neurons that scale to millions of neurons in a single state-of-the-art Field Programmable Gate Array (FPGA). We demonstrate the validity of the design methodology through the implementation of cortical development in a circuit of spiking neurons, STDP synapses, and neural architecture optimization.

Blood vessel segmentation methodologies in retinal images--a survey

Retinal vessel segmentation algorithms are a fundamental component of automatic retinal disease screening systems. This work examines the blood vessel segmentation methodologies in two dimensional retinal images acquired from a fundus camera and a survey of techniques is presented. The aim of this paper is to review, analyze and categorize the retinal vessel extraction algorithms, techniques and methodologies, giving a brief description, highlighting the key points and the performance measures. We intend to give the reader a framework for the existing research; to introduce the range of retinal vessel segmentation algorithms; to discuss the current trends and future directions and summarize the open problems. The performance of algorithms is compared and analyzed on two publicly available databases (DRIVE and STARE) of retinal images using a number of measures which include accuracy, true positive rate, false positive rate, sensitivity, specificity and area under receiver operating characteristic (ROC) curve.

CMOS REALIZATION OF A 2-LAYER CNN UNIVERSAL MACHINE CHIP

Some features of the biological retina can be modelled by a 2-layer cellular neural network (CNN) composed of locally connected elementary nonlinear processors. In order to explore these complex spatiotemporal dynamics for image processing, a prototype chip has been designed and fabricated in a 0.5μm CMOS technology. Design challenges, trade-offs, the building blocks and the tests results for this system with 0.5×106 transistors, most of them operating in analog mode, are presented in this paper.

Cellular neural networks, the Navier-Stokes equation, and microarray image reconstruction

Although the last decade has witnessed a great deal of improvements achieved for the microarray technology, many major developments in all the main stages of this technology, including image processing, are still needed. Some hardware implementations of microarray image processing have been proposed in the literature and proved to be promising alternatives to the currently available software systems. However, the main drawback of those proposed approaches is the unsuitable addressing of the quantification of the gene spot in a realistic way without any assumption about the image surface. Our aim in this paper is to present a new image-reconstruction algorithm using the cellular neural network that solves the Navier-Stokes equation. This algorithm offers a robust method for estimating the background signal within the gene-spot region. The MATCNN toolbox for Matlab is used to test the proposed method. Quantitative comparisons are carried out, i.e., in terms of objective criteria, between our approach and some other available methods. It is shown that the proposed algorithm gives highly accurate and realistic measurements in a fully automated manner within a remarkably efficient time.

Evolving mazes from images

We propose a novel reaction diffusion (RD) simulator to evolve image-resembling mazes. The evolved mazes faithfully preserve the salient interior structures in the source images. Since it is difficult to control the generation of desired patterns with traditional reaction diffusion, we develop our RD simulator on a different computational platform, cellular neural networks. Based on the proposed simulator, we can generate the mazes that exhibit both regular and organic appearance, with uniform and/or spatially varying passage spacing. Our simulator also provides high controllability of maze appearance. Users can directly and intuitively "paint" to modify the appearance of mazes in a spatially varying manner via a set of brushes. In addition, the evolutionary nature of our method naturally generates maze without any obvious seam even though the input image is a composite of multiple sources. The final maze is obtained by determining a solution path that follows the user-specified guiding curve. We validate our method by evolving several interesting mazes from different source images.

Evaluation of the traffic parameters in a metropolitan area by fusing visual perceptions and CNN processing of webcam images

This paper proposes a traffic monitoring architecture based on a high-speed communication network whose nodes are equipped with fuzzy processors and cellular neural network (CNN) embedded systems. It implements a real-time mobility information system where visual human perceptions sent by people working on the territory and video-sequences of traffic taken from webcams are jointly processed to evaluate the fundamental traffic parameters for every street of a metropolitan area. This paper presents the whole methodology for data collection and analysis and compares the accuracy and the processing time of the proposed soft computing techniques with other existing algorithms. Moreover, this paper discusses when and why it is recommended to fuse the visual perceptions of the traffic with the automated measurements taken from the webcams to compute the maximum traveling time that is likely needed to reach any destination in the traffic network.

A bio-inspired two-layer mixed-signal flexible programmable chip for early vision

A bio-inspired model for an analog programmable array processor (APAP), based on studies on the vertebrate retina, has permitted the realization of complex programmable spatio-temporal dynamics in VLSI. This model mimics the way in which images are processed in the visual pathway, what renders a feasible alternative for the implementation of early vision tasks in standard technologies. A prototype chip has been designed and fabricated in 0.5 /spl mu/m CMOS. It renders a computing power per silicon area and power consumption that is amongst the highest reported for a single chip. The details of the bio-inspired network model, the analog building block design challenges and trade-offs and some functional tests results are presented in this paper.

Image processing for a tactile/vision substitution system using digital CNN

In view of the parallel processing and easy implementation properties of CNN, we propose to use digital CNN as the image processor of a tactile/vision substitution system (TVSS). The digital CNN processor is used to execute the wavelet down-sampling filtering and the half-toning operations, aiming to extract important features from the images. A template combination method is used to embed the two image processing functions into a single CNN processor. The digital CNN processor is implemented on an intellectual property (IP) and is implemented on a XILINX VIRTEX II 2000 FPGA board. Experiments are designated to test the capability of the CNN processor in the recognition of characters and human subjects in different environments. The experiments demonstrates impressive results, which proves the proposed digital CNN processor a powerful component in the design of efficient tactile/vision substitution systems for the visually impaired people.