To address the issues of low accuracy and high false positive rate in traditional Otsu algorithm for defect detection on infrared images of wind turbine blades(WTB),this paper proposes a technique that combines morpho...To address the issues of low accuracy and high false positive rate in traditional Otsu algorithm for defect detection on infrared images of wind turbine blades(WTB),this paper proposes a technique that combines morphological image enhancement with an improved Otsu algorithm.First,mathematical morphology’s differential multi-scale white and black top-hat operations are applied to enhance the image.The algorithm employs entropy as the objective function to guide the iteration process of image enhancement,selecting appropriate structural element scales to execute differential multi-scale white and black top-hat transformations,effectively enhancing the detail features of defect regions and improving the contrast between defects and background.Afterwards,grayscale inversion is performed on the enhanced infrared defect image to better adapt to the improved Otsu algorithm.Finally,by introducing a parameter K to adjust the calculation of inter-class variance in the Otsu method,the weight of the target pixels is increased.Combined with the adaptive iterative threshold algorithm,the threshold selection process is further fine-tuned.Experimental results show that compared to traditional Otsu algorithms and other improvements,the proposed method has significant advantages in terms of defect detection accuracy and reducing false positive rates.The average defect detection rate approaches 1,and the average Hausdorff distance decreases to 0.825,indicating strong robustness and accuracy of the method.展开更多
Matrix completion is the extension of compressed sensing.In compressed sensing,we solve the underdetermined equations using sparsity prior of the unknown signals.However,in matrix completion,we solve the underdetermin...Matrix completion is the extension of compressed sensing.In compressed sensing,we solve the underdetermined equations using sparsity prior of the unknown signals.However,in matrix completion,we solve the underdetermined equations based on sparsity prior in singular values set of the unknown matrix,which also calls low-rank prior of the unknown matrix.This paper firstly introduces basic concept of matrix completion,analyses the matrix suitably used in matrix completion,and shows that such matrix should satisfy two conditions:low rank and incoherence property.Then the paper provides three reconstruction algorithms commonly used in matrix completion:singular value thresholding algorithm,singular value projection,and atomic decomposition for minimum rank approximation,puts forward their shortcoming to know the rank of original matrix.The Projected Gradient Descent based on Soft Thresholding(STPGD),proposed in this paper predicts the rank of unknown matrix using soft thresholding,and iteratives based on projected gradient descent,thus it could estimate the rank of unknown matrix exactly with low computational complexity,this is verified by numerical experiments.We also analyze the convergence and computational complexity of the STPGD algorithm,point out this algorithm is guaranteed to converge,and analyse the number of iterations needed to reach reconstruction error.Compared the computational complexity of the STPGD algorithm to other algorithms,we draw the conclusion that the STPGD algorithm not only reduces the computational complexity,but also improves the precision of the reconstruction solution.展开更多
为提高快速迭代收缩阈值算法(Fast Iterative Shrinkage-Thresholding Algorithm,FISTA)在反卷积波束形成中的空间分辨率以及计算速度,采用基于快速傅里叶变换的声学模型,引入过松弛方法和“贪婪”重启策略,提出两种改进的快速迭代收缩...为提高快速迭代收缩阈值算法(Fast Iterative Shrinkage-Thresholding Algorithm,FISTA)在反卷积波束形成中的空间分辨率以及计算速度,采用基于快速傅里叶变换的声学模型,引入过松弛方法和“贪婪”重启策略,提出两种改进的快速迭代收缩阈值算法,即基于快速傅里叶变换的过松弛单调快速迭代收缩阈值算法(Over-relaxed Monotone Fast Iterative Shrinkage-Thresholding Algorithm based on Fast Fourier Transform,FFT-OMFISTA)和基于快速傅里叶变换的“贪婪”快速迭代收缩阈值算法("Greedy"Fast Iterative Shrinkage-Thresholding Algorithm based on Fast Fourier Transform,FFT-GFISTA),并应用于反卷积波束形成的求解过程中。设计了单声源和双声源的仿真与实验,验证了所提算法的有效性与优越性。结果表明,两种所提算法都具有良好的性能,都能在声源定位中实现更高的空间分辨率以及更快的计算速度。展开更多
The iterative hard thresholding(IHT)algorithm is a powerful and efficient algorithm for solving l_(0)-regularized problems and inspired many applications in sparse-approximation and image-processing fields.Recently,so...The iterative hard thresholding(IHT)algorithm is a powerful and efficient algorithm for solving l_(0)-regularized problems and inspired many applications in sparse-approximation and image-processing fields.Recently,some convergence results are established for the proximal scheme of IHT,namely proximal iterative hard thresholding(PIHT)algorithm(Blumensath and Davies,in J Fourier Anal Appl 14:629–654,2008;Hu et al.,Methods 67:294–303,2015;Lu,Math Program 147:125–154,2014;Trzasko et al.,IEEE/SP 14th Workshop on Statistical Signal Processing,2007)on solving the related l_(0)-optimization problems.However,the complexity analysis for the PIHT algorithm is not well explored.In this paper,we aim to provide some complexity estimations for the PIHT sequences.In particular,we show that the complexity of the sequential iterate error is at o(1/k).Under the assumption that the objective function is composed of a quadratic convex function and l_(0)regularization,we show that the PIHT algorithm has R-linear convergence rate.Finally,we illustrate some applications of this algorithm for compressive sensing reconstruction and sparse learning and validate the estimated error bounds.展开更多
针对传统波达方向(Direction of Arrival, DOA)估计算法在低信噪比、小快拍的条件下估计精度不高的问题,提出了一种基于迭代收缩阈值算法的矢量水听器阵列多快拍DOA估计方法。首先,对空域进行等角度划分,构造超完备冗余字典,建立基于信...针对传统波达方向(Direction of Arrival, DOA)估计算法在低信噪比、小快拍的条件下估计精度不高的问题,提出了一种基于迭代收缩阈值算法的矢量水听器阵列多快拍DOA估计方法。首先,对空域进行等角度划分,构造超完备冗余字典,建立基于信号多快拍条件下的DOA估计模型,然后,采用迭代收缩阈值算法解决稀疏重构问题,求解出信号的稀疏系数矩阵,最后,将稀疏矩阵中行向量的范数映射到划分好的网格上,得到DOA估计值。仿真实验结果表明:该方法在低信噪比、小快拍条件下比OMP、 MUSIC和CBF等传统算法拥有更高的DOA估计精度和更强的鲁棒性。展开更多
在时分双工(TDD)毫米波大规模多输入多输出(MIMO)系统中,因为波束空间信道具有稀疏性,导致将低维测量数据重建为原始高维信道时会带来较高的复杂度。针对上行链路,在不考虑稀疏度的情况下,将传统优化算法和基于数据驱动的深度学习方法...在时分双工(TDD)毫米波大规模多输入多输出(MIMO)系统中,因为波束空间信道具有稀疏性,导致将低维测量数据重建为原始高维信道时会带来较高的复杂度。针对上行链路,在不考虑稀疏度的情况下,将传统优化算法和基于数据驱动的深度学习方法相结合,提出一种改进的基于深度学习的波束空间信道估计算法。从重建过程入手,通过交替建立梯度下降模块(GDM)和近端映射模块(PMM)来构建网络。首先根据SalehValenzuela信道模型进行理论公式推导并生成信道数据;其次构建一个由传统迭代收缩阈值算法(ISTA)的更新步骤所展开的多层网络,并将数据传输到该网络,每层对应于一次类似ISTA的迭代;最后对训练好的模型进行在线测试,恢复出待估计的信道。构建Py Torch环境,将该算法与正交匹配追踪(OMP)算法、近似消息传递(AMP)算法、可学习的近似消息传递(LAMP)算法、高斯混合LAMP(GM-LAMP)算法进行对比,结果表明:在估计精度方面,所提算法相对表现较好的深度学习算法LAMP、GM-LAMP分别提升约3.07和2.61 d B,较传统算法OMP、AMP分别提升约11.12和9.57 d B;在参数量方面,所提算法较LAMP、GM-LAMP分别减少约39%和69%。展开更多
基金supported by Natural Science Foundation of Jilin Province(YDZJ202401352ZYTS).
文摘To address the issues of low accuracy and high false positive rate in traditional Otsu algorithm for defect detection on infrared images of wind turbine blades(WTB),this paper proposes a technique that combines morphological image enhancement with an improved Otsu algorithm.First,mathematical morphology’s differential multi-scale white and black top-hat operations are applied to enhance the image.The algorithm employs entropy as the objective function to guide the iteration process of image enhancement,selecting appropriate structural element scales to execute differential multi-scale white and black top-hat transformations,effectively enhancing the detail features of defect regions and improving the contrast between defects and background.Afterwards,grayscale inversion is performed on the enhanced infrared defect image to better adapt to the improved Otsu algorithm.Finally,by introducing a parameter K to adjust the calculation of inter-class variance in the Otsu method,the weight of the target pixels is increased.Combined with the adaptive iterative threshold algorithm,the threshold selection process is further fine-tuned.Experimental results show that compared to traditional Otsu algorithms and other improvements,the proposed method has significant advantages in terms of defect detection accuracy and reducing false positive rates.The average defect detection rate approaches 1,and the average Hausdorff distance decreases to 0.825,indicating strong robustness and accuracy of the method.
基金Supported by the National Natural Science Foundation ofChina(No.61271240)Jiangsu Province Natural Science Fund Project(No.BK2010077)Subject of Twelfth Five Years Plans in Jiangsu Second Normal University(No.417103)
文摘Matrix completion is the extension of compressed sensing.In compressed sensing,we solve the underdetermined equations using sparsity prior of the unknown signals.However,in matrix completion,we solve the underdetermined equations based on sparsity prior in singular values set of the unknown matrix,which also calls low-rank prior of the unknown matrix.This paper firstly introduces basic concept of matrix completion,analyses the matrix suitably used in matrix completion,and shows that such matrix should satisfy two conditions:low rank and incoherence property.Then the paper provides three reconstruction algorithms commonly used in matrix completion:singular value thresholding algorithm,singular value projection,and atomic decomposition for minimum rank approximation,puts forward their shortcoming to know the rank of original matrix.The Projected Gradient Descent based on Soft Thresholding(STPGD),proposed in this paper predicts the rank of unknown matrix using soft thresholding,and iteratives based on projected gradient descent,thus it could estimate the rank of unknown matrix exactly with low computational complexity,this is verified by numerical experiments.We also analyze the convergence and computational complexity of the STPGD algorithm,point out this algorithm is guaranteed to converge,and analyse the number of iterations needed to reach reconstruction error.Compared the computational complexity of the STPGD algorithm to other algorithms,we draw the conclusion that the STPGD algorithm not only reduces the computational complexity,but also improves the precision of the reconstruction solution.
文摘为提高快速迭代收缩阈值算法(Fast Iterative Shrinkage-Thresholding Algorithm,FISTA)在反卷积波束形成中的空间分辨率以及计算速度,采用基于快速傅里叶变换的声学模型,引入过松弛方法和“贪婪”重启策略,提出两种改进的快速迭代收缩阈值算法,即基于快速傅里叶变换的过松弛单调快速迭代收缩阈值算法(Over-relaxed Monotone Fast Iterative Shrinkage-Thresholding Algorithm based on Fast Fourier Transform,FFT-OMFISTA)和基于快速傅里叶变换的“贪婪”快速迭代收缩阈值算法("Greedy"Fast Iterative Shrinkage-Thresholding Algorithm based on Fast Fourier Transform,FFT-GFISTA),并应用于反卷积波束形成的求解过程中。设计了单声源和双声源的仿真与实验,验证了所提算法的有效性与优越性。结果表明,两种所提算法都具有良好的性能,都能在声源定位中实现更高的空间分辨率以及更快的计算速度。
基金supported by the National Natural Science Foundation of China(No.91330102)973 program(No.2015CB856000).
文摘The iterative hard thresholding(IHT)algorithm is a powerful and efficient algorithm for solving l_(0)-regularized problems and inspired many applications in sparse-approximation and image-processing fields.Recently,some convergence results are established for the proximal scheme of IHT,namely proximal iterative hard thresholding(PIHT)algorithm(Blumensath and Davies,in J Fourier Anal Appl 14:629–654,2008;Hu et al.,Methods 67:294–303,2015;Lu,Math Program 147:125–154,2014;Trzasko et al.,IEEE/SP 14th Workshop on Statistical Signal Processing,2007)on solving the related l_(0)-optimization problems.However,the complexity analysis for the PIHT algorithm is not well explored.In this paper,we aim to provide some complexity estimations for the PIHT sequences.In particular,we show that the complexity of the sequential iterate error is at o(1/k).Under the assumption that the objective function is composed of a quadratic convex function and l_(0)regularization,we show that the PIHT algorithm has R-linear convergence rate.Finally,we illustrate some applications of this algorithm for compressive sensing reconstruction and sparse learning and validate the estimated error bounds.
文摘针对传统波达方向(Direction of Arrival, DOA)估计算法在低信噪比、小快拍的条件下估计精度不高的问题,提出了一种基于迭代收缩阈值算法的矢量水听器阵列多快拍DOA估计方法。首先,对空域进行等角度划分,构造超完备冗余字典,建立基于信号多快拍条件下的DOA估计模型,然后,采用迭代收缩阈值算法解决稀疏重构问题,求解出信号的稀疏系数矩阵,最后,将稀疏矩阵中行向量的范数映射到划分好的网格上,得到DOA估计值。仿真实验结果表明:该方法在低信噪比、小快拍条件下比OMP、 MUSIC和CBF等传统算法拥有更高的DOA估计精度和更强的鲁棒性。
文摘在时分双工(TDD)毫米波大规模多输入多输出(MIMO)系统中,因为波束空间信道具有稀疏性,导致将低维测量数据重建为原始高维信道时会带来较高的复杂度。针对上行链路,在不考虑稀疏度的情况下,将传统优化算法和基于数据驱动的深度学习方法相结合,提出一种改进的基于深度学习的波束空间信道估计算法。从重建过程入手,通过交替建立梯度下降模块(GDM)和近端映射模块(PMM)来构建网络。首先根据SalehValenzuela信道模型进行理论公式推导并生成信道数据;其次构建一个由传统迭代收缩阈值算法(ISTA)的更新步骤所展开的多层网络,并将数据传输到该网络,每层对应于一次类似ISTA的迭代;最后对训练好的模型进行在线测试,恢复出待估计的信道。构建Py Torch环境,将该算法与正交匹配追踪(OMP)算法、近似消息传递(AMP)算法、可学习的近似消息传递(LAMP)算法、高斯混合LAMP(GM-LAMP)算法进行对比,结果表明:在估计精度方面,所提算法相对表现较好的深度学习算法LAMP、GM-LAMP分别提升约3.07和2.61 d B,较传统算法OMP、AMP分别提升约11.12和9.57 d B;在参数量方面,所提算法较LAMP、GM-LAMP分别减少约39%和69%。
