Victor Barroso · 1999
The super-exponential algorithm is a block-based tech-nique for blind channel equalization and system identi-cation. Due to its fast convergence rate, and no a pri-ori parameterization other than the block length, it is a useful tool for linear equalization of moderately distortive channels. This paper presents a recursive implementation of the super-exponential algorithm for fractionally-sampled PAM signals. Although the resulting algorithm is still block-based, recursive propagation of several key variables allows the block length to be signicantly reduced without compromising the algorithm's accuracy or speed, thereby enhancing its ability to track channel variations. The con-vergence rate is only mildly in\nuenced by specic channel responses, and oversampling provides smaller output vari-ance and almost perfect tolerance to sampling errors. Simu-lation results demonstrate the eectiveness of the proposed technique. 1.
Cited by 0
João Gomes, V. Barroso · 1999
The super-exponential algorithm is a block-based technique for blind channel equalization and system identification. Due to its fast convergence rate, and no a priori parameterization other than the block length, it is a useful tool for linear equalization of moderately distortive channels. This paper presents a recursive implementation of the super-exponential algorithm for fractionally-sampled PAM signals. Although the resulting algorithm is still block-based, recursive propagation of several key variables allows the block length to be significantly reduced without compromising the algorithm's accuracy or speed, thereby enhancing its ability to track channel variations. The convergence rate is only mildly influenced by specific channel responses, and oversampling provides smaller output variance and almost perfect tolerance to sampling errors. Simulation results demonstrate the effectiveness of the proposed technique.
Cited by 2
Kai‐Bor Yu · IEE Proceedings - Radar Sonar and Navigation · 1994
Conventional methods of tracking a target using monopulse radar are severely limited in the presence of multipath, particularly at low-elevation angles. A technique using recursive eigendecomposition together with frequency-agile waveforms for tracking low-elevation targets in multipath has been developed. Frequency agility decorrelates the coherence of the direct and specular components on a pulse-to-pulse basis. This frequency agility also provides some robustness to the severe signal cancellation that arises when the direct and specular components are 180 degrees out of phase. The proposed high-resolution, eigenstructure-based algorithm for solving the low-angle tracking problem can be implemented in three steps: (i) updating the covariance matrix, (ii) updating the eigenvalue decomposition of the covariance matrix, and (iii) updating the angle estimates by searching for the peaks of the updated MUSIC spectrum, or solving the zeros derived by the minimum-norm polynomial coefficient. In addition, a phased-array implementation using three orthogonal simultaneous beams placed at half null-beamwidth apart has been developed. The angles can be determined by explicit computation or by a calibration curve in a lookup table such as in monopulse processing for angle estimation using two simultaneous beams.
Cited by 12
Xiaowei Han, Lei Wang, Xiaopeng Wang et al. · Sensors · 2023
In recent years, deep convolutional neural networks (CNNs) have made significant progress in single-image super-resolution (SISR) tasks. Despite their good performance, the single-image super-resolution task remains a challenging one due to problems with underutilization of feature information and loss of feature details. In this paper, a multi-scale recursive attention feature fusion network (MSRAFFN) is proposed for this purpose. The network consists of three parts: a shallow feature extraction module, a multi-scale recursive attention feature fusion module, and a reconstruction module. The shallow features of the image are first extracted by the shallow feature extraction module. Then, the feature information at different scales is extracted by the multi-scale recursive attention feature fusion network block (MSRAFFB) to enhance the channel features of the network through the attention mechanism and fully fuse the feature information at different scales in order to improve the network's performance. In addition, the image features at different levels are integrated through cross-layer connections using residual connections. Finally, in the reconstruction module, the upsampling capability of the deconvolution module is used to enlarge the image while extracting its high-frequency information in order to obtain a sharper high-resolution image and achieve a better visual effect. Through extensive experiments on a benchmark dataset, the proposed network model is shown to have better performance than other models in terms of both subjective visual effects and objective evaluation metrics.
Cited by 8Open access
Haixia Liu, Mingliang Wang · PLoS ONE · 2025
Image super-resolution reconstruction is one of the important application branches of computer vision in many fields. However, in actual complex environments, images are often subject to various interferences, leading to severe distortion in the reconstructed images. To address this issue, this study innovatively combines multi-scale feature extraction (MSFE) and attention feature fusion (AFF). After optimization, the multi-scale recursive attentional feature fusion block (MSRAFFB) and MSRAFFB Network (MSRAFFB-Net) application algorithms are proposed. Simulation experiments on standard datasets demonstrated that MSRAFFB was crucial for enhancing the overall performance of the algorithm, improving reconstruction quality over the baseline. Additionally, through reasonable network structure design (such as increasing module depth and branch complexity), MSRAFFB-Net effectively reduced reconstruction error and improved perceptual quality. The algorithm exhibited high reconstruction accuracy across different magnification factors. Furthermore, to assess the algorithm's performance in more realistic and complex environments, the study conducted performance experiments in actual application scenarios. The results indicate that the algorithm can effectively improve the resolution and visual quality of reconstructed images while preserving the original image's characteristic information. In summary, the proposed algorithm significantly enhances the accuracy and robustness of image reconstruction, holding positive implications for advancing the application of computer vision technology in various real-world scenarios requiring high-resolution image processing.
Cited by 0Open access
Pasquale Cascarano, Maria Colomba Comes, Arianna Mencattini et al. · Medical Image Analysis · 2021
Cited by 1Open access
Pasquale Cascarano, Maria Colomba Comes, Arianna Mencattini et al. · Cineca Institutional Research Information System (Tor Vergata University) · 2021
Biological experiments based on organ-on-chips (OOCs) exploit light Time-Lapse Microscopy (TLM) for a direct observation of cell movement that is an observable signature of underlying biological processes. A high spatial resolution is essential to capture cell dynamics and interactions from recorded experiments by TLM. Unfortunately, due to physical and cost limitations, acquiring high resolution videos is not always possible. To overcome the problem, we present here a new deep learning-based algorithm that extends the well-known Deep Image Prior (DIP) to TLM Video Super Resolution without requiring any training. The proposed Recursive Deep Prior Video method introduces some novelties. The weights of the DIP network architecture are initialized for each of the frames according to a new recursive updating rule combined with an efficient early stopping criterion. Moreover, the DIP loss function is penalized by two different Total Variation-based terms. The method has been validated on synthetic, i.e., artificially generated, as well as real videos from OOC experiments related to tumor-immune interaction. The achieved results are compared with several state-of-the-art trained deep learning Super Resolution algorithms showing outstanding performances.
Cited by 38Open access
Pasquale Cascarano, Maria Colomba Comes, Arianna Mencattini et al. · arXiv (Cornell University) · 2020
Biological experiments based on organ-on-chips (OOCs) exploit light\nTime-Lapse Microscopy (TLM) for a direct observation of cell movement that is\nan observable signature of underlying biological processes. A high spatial\nresolution is essential to capture cell dynamics and interactions from recorded\nexperiments by TLM. Unfortunately, due to physical and cost limitations,\nacquiring high resolution videos is not always possible. To overcome the\nproblem, we present here a new deep learning-based algorithm that extends the\nwell known Deep Image Prior (DIP) to TLM Video Super Resolution (SR) without\nrequiring any training. The proposed Recursive Deep Prior Video (RDPV) method\nintroduces some novelties. The weights of the DIP network architecture are\ninitialized for each of the frames according to a new recursive updating rule\ncombined with an efficient early stopping criterion. Moreover, the DIP loss\nfunction is penalized by two different Total Variation (TV) based terms. The\nmethod has been validated on synthetic, i.e., artificially generated, as well\nas real videos from OOC experiments related to tumor-immune interaction.\nAchieved results are compared with several state-of-the-art trained deep\nlearning SR algorithms showing outstanding performances.\n
Cited by 0Open access
Samir Nassiri, Moussa Labbadi, Chakib Chatri et al. · IEEE Control Systems Letters · 2023
To ensure fast convergence to the operating point near the design point of the pump with minimal motor induction’s power losses, and to achieve high dynamical performance under various uncertainties due to the influence of the environment on the output flow, we propose a novel optimal recursive terminal sliding-mode control strategy (ORTSM) using super-twisting algorithm (STA), with introducing quantum particle swarm optimization algorithm (QPSO). The QPSO was developed for calculating optimal input parameters of the pump system. Meanwhile, by introducing a fast nonsingular terminal sliding surface related to the cost function of the Linear Quadratic Regulator (LQR), we design an ORTSM manifold to achieve not only the fast convergence, and dynamic stability, but also to minimize the input energy. Afterward, we introduce the modified STA to provide good performance in the reaching phase. Furthermore, the stability of the closed-loop control system is assured by the Lyapunov approach. Finally, a comparative simulation analysis with two existing control schemes were performed to demonstrate the superiority and effectiveness of our proposed control scheme.
Cited by 2Open access
Zhoujun Li, Yun Sun, Deqiang Wang · 2007
The cubelike recursive networks is a special sub family of the binary interconnection networks. Typical cubelike recursive networks include the hypercube, the crossed cube, the Mobius cube, the generalized twisted cube, the twisted n-cube and the twisted-cube connected network. In a general sense, lots of their topological properties and network parameters are identical, but their diameters are quite different. This work makes the following contributions: Firstly, the definitions of sub-network and super-network are introduced to explain the recursive nature on structure of the cubelike recursive networks. Secondly, the supremum and infimum of the cubelike recursive networks' diameters are n and [~(n +1)/2] respectively, which are proved according to these definitions. Finally, a routing algorithm of cubelike recursive networks is proposed, with an example presented to explain how the algorithm works.
Cited by 0
Jesna Anver, P Abdulla · IET Image Processing · 2019
The online kernel adaptive filters are non‐linear filters which provide impulse response and are more efficient compared to other kernel algorithms. The performance of kernel adaptive filters depends on dictionary size. Here the single‐image super‐resolution using online kernel adaptive filters is a learning‐based method. The algorithm generates a sparser solution for obtaining high‐resolution image from a low‐resolution image. It finds out a dictionary with most significant set of basis vectors using the spatial similarity among the dictionaries created from the low‐resolution and high‐resolution image patches in the training set. The dictionary is utilised to generate the high‐resolution image. The algorithm is analysed on three different kernel adaptive filters, extended kernel recursive least squares, kernel recursive least squares tracker and naive online regularised risk minimisation algorithm. The performance of the super‐resolution method is evaluated on a large number of images and is compared with the state‐of‐the art non‐linear solutions to the super‐resolution. The results show a better progress in peak signal‐to‐noise ratio up to 1.2 dB.
Cited by 1
Roumiana Ilieva, Kiril Anguelov, Yoto Nikolov · AIP conference proceedings · 2019
Speaking for artificial intelligence and mathematics, some will say that they are two different fields. But we can assure you that mathematical models are its foundations, especially the main subset machine learning. Investigating these models and algorithms, we have the possibility to predict the abilities and limitations of artificial intelligence. Utilization of super-recursive algorithms for proving theorems, makes many such systems complete and decidable. Here we will discuss the main differences of recursive and super-recursive algorithm in particular Turing machine and inductive Turing machine. Our main statement is that artificial intelligent based on inductive Turing machine is much more powerful and will greatly enhance machine learning capabilities. As theorem proving and logic used commonly in artificial intelligence, these results will have a very important role in its implications.
Cited by 9Open access
Melba M. Crawford, Patrick Lau · 1993
A new edge detection algorithm has been developed and implemented that has both good speed and accuracy properties. The accuracy of the approach is derived from scale adaptation through anisotropic diffusion. The speed of the filter is based on recursive filtering. Specifically, the algorithm is implemented through a decomposition of the recursive filter into a convolution of two independent half space filters. This further improves the speed of the recursive filter and facilitates scale adaptation. The resulting algorithm is 1 - 2 orders of magnitude faster than comparable Gaussian based frequency domain and spatial domain methods. The new filter is omni-directional and super-elongated. It is also contour following, has computational complexity which is independent of scale, and has no truncation noise. The algorithm has been implemented and successfully applied to SPOT XS and Landsat MSS and TM imagery as one component of a region based image segmentation scheme.
Cited by 2
Ghassan Z. Qadah · [1990] Proceedings of the 2nd International IEEE Conference on Tools for Artificial Intelligence · 2002
A generic algorithm is presented, suitable for processing an important class of recursive queries, the so-called instantiated transitive-closure (TC) queries. The most important characteristics of this algorithm is that it reads any data-page from the system's disk at most once, i.e., the worst-case I/O behavior of this algorithm is linear with respect to the number of pages storing the database. Several variants, each with a different main-memory requirement, to the super-TC algorithm as well as a comparative performance evaluation of these variants are presented. The super-TC variant with minimum main-memory requirement is determined. A comparison of this variant with the more traditional delta -wavefront algorithm reveals the superiority of the super-TC variant, which is up to seven times faster.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
Cited by 0
Sheng Ren, Jianqi Li, Tianyi Tu et al. · Security and Communication Networks · 2021
Video surveillance plays an increasingly important role in public security and is a technical foundation for constructing safe and smart cities. The traditional video surveillance systems can only provide real-time monitoring or manually analyze cases by reviewing the surveillance video. So, it is difficult to use the data sampled from the surveillance video effectively. In this paper, we proposed an efficient video detection object super-resolution with a deep fusion network for public security. Firstly, we designed a super-resolution framework for video detection objects. By fusing object detection algorithms, video keyframe selection algorithms, and super-resolution reconstruction algorithms, we proposed a deep learning-based intelligent video detection object super-resolution (SR) method. Secondly, we designed a regression-based object detection algorithm and a key video frame selection algorithm. The object detection algorithm is used to assist police and security personnel to track suspicious objects in real time. The keyframe selection algorithm can select key information from a large amount of redundant information, which helps to improve the efficiency of video content analysis and reduce labor costs. Finally, we designed an asymmetric depth recursive back-projection network for super-resolution reconstruction. By combining the advantages of the pixel-based super-resolution algorithm and the feature space-based super-resolution algorithm, we improved the resolution and the visual perception clarity of the key objects. Extensive experimental evaluations show the efficiency and effectiveness of our method.
Cited by 11Open access
César Morales De Jesús, Jorge Antonio Torres Muñoz, Jesús Norberto Guerrero Tavarez et al. · 2025
This paper presents a robust trajectory tracking solution for Hybrid Underwater Gliders (HUGs) through a novel Backstepping-based Super-Twisting Algorithm (BSTA) control scheme. Addressing the challenges of nonlinear dynamics, environmental disturbances, and actuator constraints inherent in HUG operation, the proposed method combines the recursive design of backstepping control with the finite-time convergence and disturbance rejection properties of super-twisting sliding mode control. Lyapunov-based stability analysis guarantees closed-loop system performance under dynamic uncertainties. Comparative simulations demonstrate the controller's superiority over conventional PD and non-linear PD approaches in two operational scenarios.
Cited by 0
Zongping Li, Yi Ai, Qianfan Zhang · 2014
With the expansion and complication of supply chain networks, reliability determines the stability and sustainability of the entire supply chain. The supply chain network reliability has become one of the hard problems to solve by the traditional network theory. Yet, the super-network theory provides strong support for solving such a problem. This paper utilized the super-network theory to study the supply chain network reliability. The basic theory of the supply chain super-network was analyzed first, based on which the supply chain super-network reliability analysis model has been constructed to make the reliability problem hierarchical and systematic. Finally, a decomposition algorithm for network reliability was designed for the model, and through the recursive decomposition method, the amount of calculation has been reduced.
Cited by 0
Bee Eng Toh, Des McLernon, I. Lakkis · Electronics Letters · 1996
A method for efficiently implementing super exponential (SE) blind equalisation is proposed. The method, based on fast Kalman filtering theory, is recursive in order and in time, and leads to a significant reduction in the number of arithmetic operations. Using the order update by partitioning the covariance matrix of the first hundred data in a specific form, a fast initialisation is implemented. The resulting algorithm achieves the same theoretical fast convergence characteristics as the original SE algorithm but with a significant reduction in arithmetic operations.
Cited by 0
Assia Merzoug, Adda Ali‐Pacha, Naïma Hadj-Said et al. · Malaysian Journal of Computing and Applied Mathematics · 2019
One shows that we can build a long recurring sequence super-increasing and use it in the cryptographic system based on the Knapsack problem, for example the Merkle-Hellman cipher. Thus, we reduce the size of the key with the same system safety. For this, we have modified the generalized Fibonacci sequence to produce of a super-increasing sequence. This modification was based essentially on the use of real coefficients in the main recursive equation. The result of this, it is the transformation of the public key of the Merkle-Hellman crypto system into a secret algorithm and, for an equivalent complexity.
Cited by 0Open access
Bee Eng Toh, Des McLernon · 2002
Although research into blind equalisation algorithms has been an on-going activity for more than a decade, the existing algorithms are often inefficient in combating the impairments introduced by rapidly changing communication channels. New and efficient algorithms are thus needed. In this paper, an efficient blind equalisation algorithm is proposed by modifying the existing super exponential algorithm. This method, based on fast Kalman filtering theory, is recursive in order and in time, and leads to a significant reduction in the number of arithmetic operations. Using the order update by partitioning the covariance matrix of the first hundred data in a specific form, fast initialisation is also implemented. The proposed algorithm requires computational complexity proportional to N but acquires the fast convergence characteristics of the super exponential algorithm. Simulation results indicate that this equaliser performs better with a soft decision device applied at the output of the equaliser for various modulation schemes and channel conditions.
Cited by 1