Search

In this paper, we propose a lossless encryption method based on dual images. A cosine perturbation is added to the generalized voltage-controlled memristor, thus constituting a generalized voltage-controlled cosine memristor. And the characteristics of the memristor are analyzed and introduced into a four-dimensional cellular neural network, thus constituting the memristor-based four-dimensional cellular neural network (M4DCNN). The parameters of the M4DCNN are adjusted to achieve a hyperchaotic state for the diffusion and scrambling stages of the image. On the key, the partial key is obtained by K-means algorithm to make the key more random. In the scrambling phase, perturbed logistic chaotic map is introduced in this paper, which makes the image pixels better satisfy the avalanche effect. Through security analysis and comparison with other encryption methods, it is found that the encryption method shows to have good encryption effect in various indexes such as key sensitivity, key space, pixel correlation, information entropy, resistance to shear attack, resistance to noise attack, UACI and NPCR analysis. [ABSTRACT FROM AUTHOR]

The purpose of this research is to investigate the synchronization and control for uncertain memristor-based cellular neural network and its application in secure communication. To address the issue, a novel sliding mode function is designed, on which the system states can effectively converge to the equilibrium point after reaching the sliding mode surface. Moreover, the corresponding controller is constructed by employing the proposed sliding mode function. The proposed control strategy achieves the synchronization of the uncertain memristor-based cellular neural network, and effectively addresses the integral saturation existing in traditional one. In addition, the control performance, including convergence speed, control accuracy, robustness and security, are significantly enhanced. Furthermore, the stability of the system is discussed based on Lyapunov theory. Finally, comparative tests and application examples are presented to verify the effectiveness of the proposed scheme. [ABSTRACT FROM AUTHOR]

In the internet era, image encryption technology plays a crucial role in ensuring the security of image information. Aiming at the problem that low-dimensional chaotic encryption has a relatively small key space and is vulnerable to chosen-plain attacks, this paper proposed a novel colour image block encryption algorithm based on the cellular neural network and the Chua's chaotic system. Firstly, the plain is separated into three primary colours. And Fourier transform is applied to achieve optical transformation, generating six chaotic sequences iteratively. Secondly, the chaotic sequences were combined in pairs. And then combined with the plain to form three sequences, which were used in the diffusion process. Finally, the improved Chua's chaotic system is used to scramble the image. Both diffusion and scrambling are performed synchronously during encryption. Through the above operation, we get the final colour encrypted picture. The experimental results show that the algorithm has excellent encryption effect and can resist common attacks. [ABSTRACT FROM AUTHOR]

Due to the fact that a memristor with memory properties is an ideal electronic component for implementation of the artificial neural synaptic function, a brand-new tristable locally active memristor model is first proposed in this paper. Here, a novel four-dimensional fractional-order memristive cellular neural network (FO-MCNN) model with hidden attractors is constructed to enhance the engineering feasibility of the original CNN model and its performance. Then, its hardware circuit implementation and complicated dynamic properties are investigated on multi-simulation platforms. Subsequently, it is used toward secure communication application scenarios. Taking it as the pseudo-random number generator (PRNG), a new privacy image security scheme is designed based on the adaptive sampling rate compressive sensing (ASR-CS) model. Eventually, the simulation analysis and comparative experiments manifest that the proposed data encryption scheme possesses strong immunity against various security attack models and satisfactory compression performance. [ABSTRACT FROM AUTHOR]

In this work we propose a new mathematical model of 2D cellular neural networks (CNN) in terms of the lattice FitzHugh-Nagumo equations with boundary feedback. The model features discrete Laplacian operators and periodic boundary feedback instead of the interior-clamped or mean-field feedback. We first prove the globally dissipative dynamics of the solutions through a priori uniform estimates. In the main result we show that the 2D cellular neural networks are exponentially synchronized if the computable threshold condition is satisfied by the synaptic gap signals of pairwise boundary cells of the grid and the system parameters with a linear feedback coupling coefficient tunable in applications. [ABSTRACT FROM AUTHOR]

This study ventures into the captivating realm of visual illusion art, where the enigmatic principles of visual perception are harnessed to enhance artistic creativity. Through mathematical modeling of the visual perception process, we uncover the essence of visual illusions and their profound impact on art and design. Leveraging cellular neural networks, this research merges dynamic processes with the WC visual illusion and color space models to craft a novel visual illusion neural network model adept at reproducing the nuances of visual illusion art. Our investigation into the application of visual illusions in art design reveals a notable affinity for the “transparent” quality, achieving a 57% certainty level and embodying the art’s ethereal nature. Furthermore, we identify significant correlations between interactive effects, color coordination, design structure, visual impact, and the overarching quality of art designs, with correlation indices of 0.508, 0.487, 0.535, and 0.602, respectively. This work highlights visual illusion’s pivotal role in propelling the field of art and design forward, thereby enriching the tapestry of human experience.

This paper presents the control design framework for the hybrid synchronization (HS) and parameter identification of the 3-cell cellular neural network. The cellular neural network (CNN) of this kind has increasing practical importance but due to its strong chaotic behavior and the presence of uncertain parameters make it difficult to design a smooth control framework. Sliding mode control (SMC) is very helpful for this kind of environment where the systems are nonlinear and have uncertain parameters and bounded disturbances. However, conventional SMC offers a dangerous chattering phenomenon, which is not acceptable in this scenario. To get chattering-free control, smooth higher-order SMC formulated on the smooth super twisting algorithm (SSTA) is proposed in this article. The stability of the sliding surface is ensured by the Lyapunov stability theory. The convergence of the error system to zero yields hybrid synchronization and the unknown parameters are computed adaptively. Finally, the results of the proposed control technique are compared with the adaptive integral sliding mode control (AISMC). Numerical simulation results validate the performance of the proposed algorithm. [ABSTRACT FROM AUTHOR]

For a graph Γ , a subset S ⊆ V Γ is known to be a dominating set, if every x ∈ V Γ \ S has at least one neighbor in D. The domination number γ (Γ) is merely the size of a smallest dominating set in Γ . The strong product P r ⊠ P s of two paths P r and P s is known as the king's graph. Interestingly, the king's graph is isomorphic to the two-parametric family of cellular neural network (CNNs). In Asad et al. (Alex Eng J 66:957–977, 2023), the authors retrieved certain structural characteristics of CNNs from their minimal dominating sets. They conjectured in Problem 8.3 that the domination number of P r ⊠ P s is ⌈ r 3 ⌉ ⌈ s 3 ⌉ . This paper solves Problem 8.3 by providing a proof to the conjecture. This result, in turn, reveals interesting topological properties such as an optimal routing for this class of neural networks. [ABSTRACT FROM AUTHOR]

The paper considers a class of discrete-time cellular neural networks (DT-CNNs) obtained by applying Euler's discretization scheme to standard CNNs. Let T be the DT-CNN interconnection matrix which is defined by the feedback cloning template. The paper shows that a DT-CNN is convergent, i.e. each solution tends to an equilibrium point, when T is symmetric and, in the case where T + E n is not positive-semidefinite, the step size of Euler's discretization scheme does not exceed a given bound ( E n is the n × n unit matrix). It is shown that two relevant properties hold as a consequence of the local and space-invariant interconnecting structure of a DT-CNN, namely: (1) the bound on the step size can be easily estimated via the elements of the DT-CNN feedback cloning template only; (2) the bound is independent of the DT-CNN dimension. These two properties make DT-CNNs very effective in view of computer simulations and for the practical applications to high-dimensional processing tasks. The obtained results are proved via Lyapunov approach and LaSalle's Invariance Principle in combination with some fundamental inequalities enjoyed by the projection operator on a convex set. The results are compared with previous ones in the literature on the convergence of DT-CNNs and also with those obtained for different neural network models as the Brain-State-in-a-Box model. Finally, the results on convergence are illustrated via the application to some relevant 2D and 1D DT-CNNs for image processing tasks. [ABSTRACT FROM AUTHOR]

A new fractional-order cellular neural network (CNN) system is solved using the Adomian decomposition method (ADM) with the hyperbolic tangent activation function in this paper. The equilibrium point is analyzed in this CNN system. The dynamical behaviors are studied as well, using a phase diagram, bifurcation diagram, Lyapunov Exponent spectrum (LEs), and spectral entropy (SE) complexity algorithm. Changing the template parameters and the order values has an impact on the dynamical behaviors. The results indicate that rich dynamical properties exist in the system, such as hyperchaotic attractors, chaotic attractors, asymptotic periodic loops, complex coexisting attractors, and interesting state transition phenomena. In addition, the digital circuit implementation of this fractional-order CNN system is completed on a digital signal processing (DSP) platform, which proves the accuracy of ADM and the physical feasibility of the CNN system. The study in this paper offers a fundamental theory for the fractional-order CNN system as it applies to secure communication and image encryption. [ABSTRACT FROM AUTHOR]

A new tri-cellular neural network(CNN) system based on double memristors is constructed which used a hyperbolic tangent function instead of the conventional segmentation function in this paper. The multiple equilibrium points existing in the CNN system are analyzed. Through Lyapunov exponential spectrum, bifurcation diagram, phase diagram, SE complexity and digital circuit implementation, the rich and complex dynamical characteristics of the double-MR-based CNN system are presented. Interestingly, changing different parameters and initial values, the system has multiple coexisting attractors which include periodic-periodic attractors, periodic-chaotic attractors, and chaotic-chaotic attractors. Finally, a hardware circuit of the memristive cellular neural network is designed and built on the basis of a DSP platform to verify the implementability of the network model. The improved double-MR-based Cellular neural network system provides a theoretical foundation in other fields of application, especially for secure communications. [ABSTRACT FROM AUTHOR]

The paper presents a secure communication scheme for the optical physical layer based on a seven-dimensional (7-D) Cellular Neural Network (CNN) hyperchaotic encryption. The encryption scheme utilizes a 7-D CNN hyperchaotic system to generate a hyperchaotic sequence as the key source. A portion of this key is selected to encrypt the plaintext image. The proposed scheme has been successfully implemented in a PDM-16QAM system with a data rate of 224 Gbps over a 200 km single-mode fiber (SMF). Experimental results show that authorized users can successfully decrypt the received signal, while eavesdroppers are unable to obtain useful information with a bit error rate (BER) of approximately 0.5. The key space of the scheme is 101792. Through key sensitivity analysis and key space analysis, it is known that the proposed encryption system can effectively resist various attacks by cryptanalysts. • A super large key space optical physical layer encryption scheme with a key space of 101792, which can effectively resist brute force attacks. • A seven-dimensional CNN hyper chaotic system and permutation encryption diffusion method are used for long-distance, high-speed optical physical layer encryption. • The optical physical layer encryption system can achieve a large key space at a lower complexity cost. And experiments have shown that it has good performance. [ABSTRACT FROM AUTHOR]

Summary: High latency and power consumption are two major problems that need to be addressed in convolutional neural networks (CNN). In this paper, the convolutional layer is replaced with a discrete‐time cellular neural network (CellNN) to overcome these problems. Multiple configurations of CellNNs are trained in a framework called TensorFlow to classify objects from the CIFAR‐10 database. Effects of the number of iterations, the number of channels, batch normalization, and activation functions on the classification accuracies are presented. It is shown that TensorFlow is a tool that is capable of training discrete‐time CellNNs. Although the accuracies of the proposed networks on CIFAR‐10 are slightly lesser than the existing CNNs, with reduced parameters and multiply‐accumulates (MACs), power consumption and computation time of our networks will be less than CNNs. [ABSTRACT FROM AUTHOR]

This paper presents a novel algorithm for image denoising using an improved nonlinear diffusion PDE model and a cellular neural network (CeNN) scheme. In particular, the images corrupted with multiplicative (speckle) noise have been considered. The proposed generalized-order nonlinear diffusion (GOND) model is solved through a suitable cellular neural network (CeNN) approach. The CeNN templates act like edge-preserving filters to reduce the multiplicative noise. The present study also gives a convergence analysis of the proposed CeNN based solution scheme. Further, the proposed scheme is numerically validated on synthetic, medical, and real SAR images. The obtained results demonstrate that the proposed algorithm provides a better way to deal with speckle noise and suppresses the staircase effects. To broaden the simulation results, the proposed method is applied to images corrupted with Gaussian, Rayleigh, and gamma noise. The proposed method for SSIM values interpret 0.2dB to 0.4dB better than state-of-the-art methods and comparative results in terms of PSNR in most test cases. [ABSTRACT FROM AUTHOR]

This work presents a cellular neural network (CNN) learning tool based on the Center of Mass Algorithm (CMA), which provides a friendly graphical user interface. As original contributions of this work, the learning tool here developed for CNN features a few adaptations in CMA to improve the training process for grayscale image filtering, such as decreasing learning rates and training with multiple image sets, and comprises many graphical user interface facilities for CNN designers. The training capability of the developed tool is validated through several filter applications, among which lowpass and highpass filters with finite or infinite impulse responses. A comparative analysis is performed between the theoretical responses of the filters and the results obtained from the simulation of a CMOS analog CNN configured by the parameters determined through the learning tool. The training process was very fast for the lowpass filters and acceptable RMS pixel errors have been obtained for all examples, confirming the reliability of the learning tool. [ABSTRACT FROM AUTHOR]

In this paper, we propose a dynamic key technique based on Cellular Neural Network (CNN) for security improvement in the orthogonal frequency division multiplexing passive optical network (OFDM-PON). To enhance the encryption scheme security, a six-dimensional CNN hyperchaotic system is employed to encrypt the data. And, the keys are divided into the dynamic and static. The dynamic key is randomly extracted from a key set by incorporating the random feature of the input data. Then, the chaotic sequence generated by the dynamic key is served as the synchronous sequence for encryption. Moreover, the chaotic sequences generated by the static keys are used to resist the chosen-plaintext attacks (CPAs) and scramble the phase of QAM symbols on the frequency domain. With these processing techniques, the multi-fold data encryption can create a key space of ∼10315 to protect against the exhaustive trial. The transmission of 10-Gb/s encrypted 16-QAM-based OFDM signal is demonstrated over 20-km single mode fiber (SMF) by experiment. The results show that our proposed scheme can provide excellent confidentiality of data transmission against the CPAs and brute-force attack.

This paper introduces a contour detection scheme to detect object contours in medical images. A new PDE model is designed by including a fractional-order regularization term, making it robust against noise and maintaining the regularity of level set function (LSF) during evolution. A cellular neural network (CeNN) model is used to solve the proposed contour detection PDE. The main advantages of using the CeNN-based approach are that it wipes out the requirement of a reinitialization of level set and can be implemented efficiently on parallel chips. Finally, an experimental study is carried out, which exhibits the feasibility of the proposed approach in contour detection from a set of medical images. [ABSTRACT FROM AUTHOR]

In this work we study the synchronization of ring-structured cellular neural networks modeled by the lattice FitzHugh-Nagumo equations with boundary feedback. Through the uniform estimates of solutions and the analysis of dissipative dynamics, the synchronization of this type neural networks is proved under the condition that the boundary gap signal exceeds the adjustable threshold. [ABSTRACT FROM AUTHOR]

This paper proposes a 7-dimensional () Cellular Neural Network (CNN) based offset quadrature amplitude modulation filter bank multicarrier (OQAM-FBMC) encryption scheme for seven core fiber. The chaotic sequences generated by 7D CNN are applied to produce the masking vectors to encrypt the phase, carrier frequency, and time. In order to verify the performance of the encryption scheme, 70 Gb/s (7×10 Gb/s) encrypted OQAM-FBMC signal transmission over 2 km 7 core fiber is experimentally demonstrated. The key space of 7D CNN can reach 101575 and the scrambling degree can be maintained at 100% regardless of the number of symbols. The experimental results also show that when some keys are compromised, the system's bit error rate (BER) can still reach above 0.46, which effectively ensures the security of the system. Due to its good performance in security, the proposed scheme has important application prospects in future optical access network. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Jilin University (Science Edition) / Jilin Daxue Xuebao (Lixue Ban) is the property of Zhongguo Xue shu qi Kan (Guang Pan Ban) Dian zi Za zhi She and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Memristor is a kind of synaptic element with nanometer size and continuously variable memristance. The bridge synaptic circuit constructed by the memristor has a simple structure and precise control. In practice, because of the non-linear characteristics of memristor, it is not easy to control synaptic circuit and errors in weights appear. Therefore, a novel memristor synaptic circuit is proposed in this paper, called the dual-mode memristor bridge synaptic neural network. The proposed method can make the weights more linear by controlling the input voltages and make the outputs more linear by using symmetrical positive and negative pulses. Therefore, the proposed synaptic circuit is easier to be controlled. In this paper, the numerical simulations are conducted and verify the feasibility. Furthermore, the simulation experiments are conducted for edge extraction of grayscale birds’ images in the airport for bird recognition applied for the bird repelling applications.

In this paper, we explore the dynamical behaviors of the 1D two-grid coupled cellular neural networks. Assuming the boundary conditions of zero-flux type, the stability of the zero equilibrium is discussed by analyzing the relevant eigenvalue problem with the aid of the decoupling method, and the conditions for the occurrence of Turing instability and Hopf bifurcation at the zero equilibrium are derived. Furthermore, the approximate expressions of the bifurcating periodic solutions are also obtained by using the Hopf bifurcation theorem. Finally, numerical simulations are provided to demonstrate the theoretical results. [ABSTRACT FROM AUTHOR]

In this paper, we propose a dynamic key technique based on Cellular Neural Network (CNN) for security improvement in the orthogonal frequency division multiplexing passive optical network (OFDM-PON). To enhance the encryption scheme security, a six-dimensional CNN hyperchaotic system is employed to encrypt the data. And, the keys are divided into the dynamic and static. The dynamic key is randomly extracted from a key set by incorporating the random feature of the input data. Then, the chaotic sequence generated by the dynamic key is served as the synchronous sequence for encryption. Moreover, the chaotic sequences generated by the static keys are used to resist the chosen-plaintext attacks (CPAs) and scramble the phase of QAM symbols on the frequency domain. With these processing techniques, the multi-fold data encryption can create a key space of ∼10315 to protect against the exhaustive trial. The transmission of 10-Gb/s encrypted 16-QAM-based OFDM signal is demonstrated over 20-km single mode fiber (SMF) by experiment. The results show that our proposed scheme can provide excellent confidentiality of data transmission against the CPAs and brute-force attack. [ABSTRACT FROM AUTHOR]

In order to enhance the chaotic degree of cellular neural network (CNN), the memristive characteristic is combined in CNN, and a five-dimensional memristive CNN hyperchaotic system is designed. A secure communication system based on chaos synchronization is constructed by the full-dimensional state observer theory. The sliding mode control method is used to control chaos synchronization between the sending end and the receiving end and improve the robustness to the parameter uncertainties and disturbances in the system. An improved sliding mode surface is designed, and its convergence time is analyzed. Image secure communication, as an example, is given to verify the effectiveness of the proposed method. Image encryption is implemented by the hyperchaotic system, and image decryption is implemented by chaos synchronization based on the sliding mode control. Simulation experiment results show that the proposed sliding mode control method can be used to achieve the chaos synchronization of the hyperchaotic systems with parameter uncertainties and disturbances, and the proposed memristive hyperchaotic system can be applied to the secure communication based on the chaos synchronization. [ABSTRACT FROM AUTHOR]

In the past, several partial differential equations (PDEs) based methods have been widely studied in image denoising. While solving these methods numerically, some parameters need to be chosen manually. This paper proposes a cellular neural network (CNN) based computational scheme for solving the nonlinear diffusion equation modeled for removing additive noise of digital images. The diffusion acts like smoothing on the noisy image, which is taken as an initial condition for the nonlinear PDE. In the proposed scheme, the template matrices of CNN evolve during the iterative diffusion and act as edge-preserving filters on the noisy images. The evolving diffusion ensures convergence of the diffusion process after a specific diffusion time. Therefore, the advantages of such a CNN-based solution scheme are more accurate restoration in terms of image quality with low computation and memory requirements. The experimental results show the effectiveness of the proposed algorithm on different sets of benchmark images degraded with additive noise. [ABSTRACT FROM AUTHOR]

Fully memristive neuron learning framework, which uses drift and diffusion memristor models to build an artificial neuron structure, becomes a hot topic recently with the development of memristor. However, some other devices like resistor or capacitor are still necessary in recent works of fully memristive learning framework. Theoretically, if one neuron is built by memristors only, the technique process will be simpler and learning framework will be more like biological brain. In this paper, a fully memristive spiking-neuron learning framework is introduced, in which a neuron structure is just built of one drift and one diffusion memristive models and spikes are used as transmission signals. The learning framework and spiking coding mode are simple and direct without any complicated calculation on hardware. To verify its merits, a feedforward neural network for pattern recognition and a cellular neural network for edge detection are designed. Experimental results show that compared to other memristive neural networks, processing speed of the proposed framework is very high, and the hardware resource is saved in pattern recognition. Further, due to the dynamic filtering function of diffusion memristor model in our learning framework, its peak signal noise ratio (PSNR) is much higher than traditional algorithms in edge detection. [ABSTRACT FROM AUTHOR]

• Topographic architectures enable fast detection of hierarchical layers. • Detection of hierarchical layers is carried out by morphological operations. • Grouping is simplified to the diffusion template of Cellular Neural Networks. • Real-time processing is achieved when implementing on topographic architectures. Information patterns are ubiquitous and their automatic detection is fundamental in many computer vision tasks. Robust and fast detection of object candidates is essential as well as the ability to adapt the system to new situations. We propose a general, task independent method that i) locates interesting patterns on binary images by creating a hierarchical layered structure based on neighborhood topography of connected components, and ii) identifies object groups applying saliency principles. Saliency values are assigned to both object groups and individuals, that can be referred as a priority queue of image regions, or alternatively a proposal for blob processing order. Empirical analysis through several computer vision examples is provided including applications for blind and visually impaired people. The proposed algorithm can be efficiently implemented on processor arrays, since it mostly contains standard topographic instructions, and we also introduce a real-time implementation on the Eye-RIS/Toshiba SPS vision system. [ABSTRACT FROM AUTHOR]

We discuss the architecture, implementation and testing of a simplicial Cellular Neural Network (CNN) vector processor core aimed at vision oriented intelligent Internet-of-Things (IoT) devices. The architecture comprises a linear array of 64 processing elements (PE), each connected to a 4 neighbor clique operating on 8-bit input and state data. A 3-bit simplicial parameter, allows multilevel function approximation and extends the functionality over previously reported chips. Input data vectors are stored in two $64 \times 64 \times 8$ -bit data caches. The chip is synthesized from a custom designed ultra low voltage CMOS library and fabricated in a 55nm CMOS technology. Dynamic voltage/frequency scaling allows operation at power supplies between 0.5 and 1.2 Volts allowing for a tradeoff between speed and power. The fabricated chip achieves an overall performance of 7.05 TOPS/W at 732fps, with a dynamic energy efficiency of 12.2fJ per operation (OP) at 1.2 Volts. [ABSTRACT FROM AUTHOR]

• An improved Cellular Neural Network (CellNN) by local-neighbor binary constraints. • A feature extraction framework based on the state and output feature maps from recurrent CellNN convolutions. • A feature map compression, joint-distribution pattern fusion and a multi-resolution scheme optimized by softmax. Texture recognition is one of the most important branches in image research. This paper mainly aims to develop a new solution to address texture recognition using a Cellular Neural Network (CellNN). Firstly, it proposes an improved model of CellNN by the binary constraints of local receptive fields, and then designs a recurrent convolution framework of such a model to generate two types of texture feature maps, including state feature maps and output feature maps. In order to obtain low-dimensional features, state feature maps are further compressed by the mapping of rotation-invariant patterns and the merging of low-frequency occurrence patterns. By the statistics of joint-distribution patterns, state feature maps and output feature maps are fused together to generate the features of single resolution. Moreover, a multi-resolution feature combination scheme is also designed by the optimization of softmax & variance and concatenation of multiple features. Finally, a fully-connected neural network is trained to work as a texture recognizer. The experimental comparisons of totally 15 algorithms on five benchmark datasets show that, on the dataset whose texture-class quantity is not beyond 30, such as Brodatz, our method could always acquire the highest recognition accuracy, outperforming any other compared ones. On the big dataset with huge texture-class quantity, such as ALOT, our method could also surpass any other non-deep-learning one, such as the state-of-the-art gLBP, only slightly falling behind the best two deep-learning ones, FV-Alex and FV-VGGVD. However, in terms of time cost, our method could always outperform any deep-learning one in feature extraction stage, and also surpass any compared one except original LBP in feature matching. [ABSTRACT FROM AUTHOR]

Aroma is one of the most important factors affecting the quality of citrus fruit and its processed products. We use electronic nose (E-nose) to detect and analyze volatile components in citrus. An E-nose is an artificial intelligence system with strong independence and fast detection speed. It combines an array of gas sensors and intelligent algorithms designed to analyze gas. Moreover, it has the ability to detect and analyze volatile components. Feature extraction is the first step of sensor signal processing and plays an important role in subsequent pattern recognition. Cellular neural network (CNN) is a real-time high-speed parallel array processor and a locally connected network, which has mature applications in the field of image processing. Previous researches have shown that CNN has an outstanding impact on image feature extractio. In this paper, the traditional CNN is improved and a template for dynamic feature extraction of the E-nose response curve is proposed. In addition, we provide users with single-template and multi-template solutions which can be applied in different environments. To free up the computational power of occupancy, the effect of the single-template version of CNN is not as effective as the multi-template version, but it still has good feature extraction ability. These two solutions prove that CNN is sensitive to dynamic features. In order to make the results more representative, we choose several traditional feature extraction methods for comparison. [ABSTRACT FROM AUTHOR]

In texture classification, local binary pattern (LBP) is currently one of the most widely-concerned feature encoding models. Most existing LBP-based texture classification methods are usually limited to single-kind texture features. In fact, an across-domain fusion of LBP features with other features, such as image contours, could be another potential path to promote texture classification. To enhance the feature modeling ability of LBP-based methods, this paper firstly designs a Cellular Neural Network (CellNN) with recurrent convolutions, initially trained by a simplified simulated-annealing algorithm, to extract informative image contours. For better reliability, a new three-channel contour extractor of deep CellNNs (i. e. , dCellNNs) is proposed. This extractor contains the initially-trained CellNNs of more than three layers, and it is further optimized by fine-tuning parameters. Moreover, a new weighted-base algorithm is designed to fulfill the fusion of the multi-scale texture features by LBPs and the contour features by dCellNNs to enhance feature representation. Finally, these enhanced features are concatenated together to generate the final multi-scale features of given texture image. On texture datasets KTH, Brodatz, OTC12 and UIUC, experiment results verify that the across-domain fusion of multi-scale LBPs and dCellNNs is efficient in capturing & enhancing texture features. With moderate feature dimensionality and computational costs, it could improve texture classification, acquiring an obvious accuracy increase on previous state-of-the-art ones, e.g. , a rise of 2.58% on KTH-TIPS2b, a rise of 3.11% on Brodatz, a rise of 0.71% on OTC12 and a rise of 0.42% on UIUC. • A multi-scale across-domain feature fusion of textures and contours is proposed. • The first deep model of CellNN (dCellNNs) with two-stage training is proposed. • The three-channel texture contour extractor of dCellNNs is designed. • A weighted-base fusing is designed for the across-domain fusion of features. [ABSTRACT FROM AUTHOR]