This paper deals with alternating optimization of sensing matrix and sparsifying dictionary for compressed systems. Under the same framework proposed by J. M. Duarte-Carvajalino G. Sapiro, a novel algorithm optimal design is derived an optimized embedded. A closed-form solution to problem obtained. new measure optimizing developed solving corresponding problem. Experiments are carried out synthetic data real images, which demonstrate promising performance algorithms superiority CS system...

This paper deals with designing sensing matrix for compressive systems. Traditionally, the optimal is designed so that Gram of equivalent dictionary as close possible to a target small mutual coherence. A novel design strategy proposed, in which, unlike traditional approaches, measure considers coherence behavior well sparse representation errors signals. The defined one minimizes this and hence expected be more robust against errors. closed-form solution derived given Gram. An alternating...

This paper deals with the design of a sensing matrix along sparse recovery algorithm by utilizing probability-based prior information for compressed systems. With knowledge probability each atom dictionary being used, diagonal weighted is obtained and then designed minimizing function such that Gram equivalent as close to possible. An analytical solution corresponding derived requires low computational complexity. We also exploit this through stage propose probability-driven orthogonal...

Abstract Automatic classification of diseases in endoscopic images is essential to the improvement diagnostic performance and reduction colorectal cancer mortality. However, due ambiguous boundary between background foreground, abnormal still challenging. To tackle such a situation, an adaptive aggregation with self‐attention network (AASAN), including global branch, local fusion proposed imitating diagnosis process endoscopists. On this basis, relative position encoding (SA‐RPE) module...

This study deals with the issue of designing sensing matrix for a compressed (CS) system assuming that dictionary is given. Traditionally, measurement small mutual coherence considered to design optimal so Gram equivalent as close target possible, where not normalised. In other words, these algorithms are designed solve CS problem using an optimisation stage followed by normalisation. To achieve global solution, novel strategy proposed gradient-based method, in which measure real considered....

The sparse representation error (SRE) exists when the images are represented sparsely. SRE is particularly large in unmanned aerial vehicles (UAV) due to disturbance of harsh environment or instability its flight, which will bring more noise. In compressed sensing (CS) system, projected measurement a significant challenge recovery accuracy images. this work, new structure proposed. Following structure, lower (LSRE) achieved by eliminating groups representation. With proposed LSRE modeling,...

The block recursive least square (BRLS) dictionary learning algorithm that dealing with training data arranged in is proposed this paper. BRLS can be used to update overcomplete for sparse signal representation. Different from traditional algorithms, designed a form and the recursion developed without using matrix inversion lemma. applied synthetic real image reconstruction. Simulation results show new achieves better performance than approaches.

This paper deals with the optimization of sensing matrices and sparsifying dictionaries for compressed systems. A gradient-based method a new measurement strategy denoted as real mutual coherence is proposed. Further more, matrix optimized by minimizing an objective function in which target Gram selected Ψ Ψ, this choice has advantages to reconstruct images more accurately efficiently. Experiments are carried out results show that obtained using proposed approach outperforms existing ones...

In this paper, we discuss the design of optimal sensing matrix Φ for compressed system, where dictionary Ψ is assumed to be given. A new measure proposed obtain matrix, which unlike traditional approaches, takes sparse representation error measurements into account, and leads a more robust CS system. gradient-based algorithm derived attack optimization problem. Simulations are performed results illustrate reconstruction accuracy performance our novel approach outperforms existing ones.

Wireless capsule endoscopy (WCE) has been widely used in the detection of digestive tract diseases because its painlessness and convenience. Accurate classification WCE abnormal images is very crucial for diagnosis treatment early gastrointestinal tumors, while it remains challenging due to ambiguous boundary between lesions normal tissues. In order overcome above limitations, a three-branch effectively fused attention guided convolutional neural network (EFAG-CNN) which imitates practical...

This paper deals with dictionary learning and optimal sensing matrix design for compressed (CS) systems. An improved version of the method directions (MOD) is proposed, which can overcome problem inversion. The defined as to find those matrices that minimize a Frobenius norm-based difference between Gram equivalent identity matrix. solution set characterized, generalization existing results. A numerical algorithm derived best among set. Simulation results are carried out, show proposed...

This paper deals with the problem of sparse recovery often found in compressive sensing applications exploiting a priori knowledge. In particular, we present knowledge-aided normalized iterative hard thresholding (KA-NIHT) algorithm that exploits information about probabilities nonzero entries. We also develop strategy to update using recursive KA-NIHT (RKA-NIHT) algorithm, which results improved recovery. Simulation illustrate and compare performance proposed existing algorithms.

Dictionary learning problem has become an active topic for decades. Most existing methods train the dictionary to adapt a particular class of signals. But as number atoms is increased represent signals much more sparsely, coherence between becomes higher. According greedy and compressed sensing theories, this goes against implementation sparse coding. In paper, novel approach proposed learn that minimizes representation error according training with taken into consideration. The constrained...

We proposed a novel sensing matrix for compressed (CS) based distributed estimation (DCE) scheme. In traditional approaches design, normalization of the column vectors is independent from optimizing mutual coherence. The algorithm integrates these two processes into one framework, which gives cost function profound physical meaning. Applying this new design results in higher reconstruction accuracy and better realtime performance on channel wireless sensor network (WSN), development DCE...

Distributed channel estimation (DCE) is one of the core research topics in wireless sensor networks (WSNs). Under hypothesis that parameters can be modeled as a sparse system, DCE based on compressed sensing (CS) an effective approach to estimation. Among all existing CS-DCE schemes, every node must store matrix whose size will increase with number parameters. Hence workload each increases. To overcome this problem, paper, scheme developed Semi-tensor product (STP) relieve sensor's storage...

In this work, we present a joint sensing matrix design and recovery algorithm based on the normalized iterative hard thresholding (NIHT) for cost-effectively solving problem of sparse recovery. particular, consider both Gram gradient-based real mutual coherence (RMC) to compute matrix, so that can closely approach relaxed equiangular tight frame (ETF. By optimizing together with its column normalization, better performance be achieved. Simulations assess proposed versus other...

This paper deals with the design problem of optimal sparsifying dictionary where measurement is not directly sparse signal but disturbed by some linear operators. Similar traditional learning problem, strategy divided into two stages. The matching pursuit method used to calculate coefficients and a new algorithm based on gradient proposed train dictionary. When being applied image inpainting learnt corrupted itself process operated fully overlapped patches resulting obtained averaging...

Compressive sensing theory shows that sparse signals can be reconstructed from far less samples than those required by the classical Shannon-Nyquist Theorem. An optimized matrix for a certain class of further reduce necessary number samples. Additionally, in order to make represented as possible, dictionary optimized. In this paper, we introduce framework joint design and dictionary. A new procedure is proposed, which based on an alternative optimization sparsifying matrix. Simulation...