arXiv subjects

Rui Zeng

Publications and source records attributed to Rui Zeng.

27 records · Page 2Linked to original sources

arXiv2019

MTRNet: A Generic Scene Text Eraser

Text removal algorithms have been proposed for uni-lingual scripts with regular shapes and layouts. However, to the best of our knowledge, a generic text removal method which is able to remove all or user-specified text regions regardless of font, script, language or shape is not available. Developing such a generic text eraser for real scenes is a challenging task, since it inherits all the challenges of multi-lingual and curved text detection and inpainting. To fill this gap, we propose a mask-based text removal network (MTRNet). MTRNet is a conditional adversarial generative network (cGAN) with an auxiliary mask. The introduced auxiliary mask not only makes the cGAN a generic text eraser, but also enables stable training and early convergence on a challenging large-scale synthetic dataset, initially proposed for text detection in real scenes. What's more, MTRNet achieves state-of-the-art results on several real-world datasets including ICDAR 2013, ICDAR 2017 MLT, and CTW1500, without being explicitly trained on this data, outperforming previous state-of-the-art methods trained directly on these datasets.

cs.CV↗

arXiv2015

Kernel principal component analysis network for image classification

In order to classify the nonlinear feature with linear classifier and improve the classification accuracy, a deep learning network named kernel principal component analysis network (KPCANet) is proposed. First, mapping the data into higher space with kernel principal component analysis to make the data linearly separable. Then building a two-layer KPCANet to obtain the principal components of image. Finally, classifying the principal components with linearly classifier. Experimental results show that the proposed KPCANet is effective in face recognition, object recognition and hand-writing digits recognition, it also outperforms principal component analysis network (PCANet) generally as well. Besides, KPCANet is invariant to illumination and stable to occlusion and slight deformation.

cs.LG↗

arXiv2015

Error Gradient-based Variable-Lp Norm Constraint LMS Algorithm for Sparse System Identification

Sparse adaptive filtering has gained much attention due to its wide applicability in the field of signal processing. Among the main algorithm families, sparse norm constraint adaptive filters develop rapidly in recent years. However, when applied for system identification, most priori work in sparse norm constraint adaptive filtering suffers from the difficulty of adaptability to the sparsity of the systems to be identified. To address this problem, we propose a novel variable p-norm constraint least mean square (LMS) algorithm, which serves as a variant of the conventional Lp-LMS algorithm established for sparse system identification. The parameter p is iteratively adjusted by the gradient descent method applied to the instantaneous square error. Numerical simulations show that this new approach achieves better performance than the traditional Lp-LMS and LMS algorithms in terms of steady-state error and convergence rate.

eess.SY↗

arXiv2015

Gradient Compared Lp-LMS Algorithms for Sparse System Identification

In this paper, we propose two novel p-norm penalty least mean square (Lp-LMS) algorithms as supplements of the conventional Lp-LMS algorithm established for sparse adaptive filtering recently. A gradient comparator is employed to selectively apply the zero attractor of p-norm constraint for only those taps that have the same polarity as that of the gradient of the squared instantaneous error, which leads to the new proposed gradient compared p-norm constraint LMS algorithm (LpGC-LMS). We explain that the LpGC-LMS can achieve lower mean square error than the standard Lp-LMS algorithm theoretically and experimentally. To further improve the performance of the filter, the LpNGC-LMS algorithm is derived using a new gradient comparator which takes the sign-smoothed version of the previous one. The performance of the LpNGC-LMS is superior to that of the LpGC-LMS in theory and in simulations. Moreover, these two comparators can be easily applied to other norm constraint LMS algorithms to derive some new approaches for sparse adaptive filtering. The numerical simulation results show that the two proposed algorithms achieve better performance than the standard LMS algorithm and Lp-LMS algorithm in terms of convergence rate and steady-state behavior in sparse system identification settings.

eess.SY↗

arXiv2015

Color Image Classification via Quaternion Principal Component Analysis Network

The Principal Component Analysis Network (PCANet), which is one of the recently proposed deep learning architectures, achieves the state-of-the-art classification accuracy in various databases. However, the performance of PCANet may be degraded when dealing with color images. In this paper, a Quaternion Principal Component Analysis Network (QPCANet), which is an extension of PCANet, is proposed for color images classification. Compared to PCANet, the proposed QPCANet takes into account the spatial distribution information of color images and ensures larger amount of intra-class invariance of color images. Experiments conducted on different color image datasets such as Caltech-101, UC Merced Land Use, Georgia Tech face and CURet have revealed that the proposed QPCANet achieves higher classification accuracy than PCANet.

cs.CV↗

arXiv2015

p Norm Constraint Leaky LMS Algorithm for Sparse System Identification

This paper proposes a new leaky least mean square (leaky LMS, LLMS) algorithm in which a norm penalty is introduced to force the solution to be sparse in the application of system identification. The leaky LMS algorithm is derived because the performance ofthe standard LMS algorithm deteriorates when the input is highly correlated. However, both ofthem do not take the sparsity information into account to yield better behaviors. As a modification ofthe LLMS algorithm, the proposed algorithm, named Lp-LLMS, incorporates a p norm penalty into the cost function ofthe LLMS to obtain a shrinkage in the weight update equation, which then enhances the performance of the filter in system identification settings, especially when the impulse response is sparse. The simulation results verify that the proposed algorithm improves the performance ofthe filter in sparse system settings in the presence ofnoisy input signals.

eess.SY↗

arXiv2015

p-norm-like Constraint Leaky LMS Algorithm for Sparse System Identification

In this paper, we propose a novel leaky least mean square (leaky LMS, LLMS) algorithm which employs a p-norm-like constraint to force the solution to be sparse in the application of system identification. As an extension of the LMS algorithm which is the most widely-used adaptive filtering technique, the LLMS algorithm has been proposed for decades, due to the deteriorated performance of the standard LMS algorithm with highly correlated input. However, both ofthem do not consider the sparsity information to have better behaviors. As a sparse-aware modification of the LLMS, our proposed Lplike-LLMS algorithm, incorporates a p-norm-like penalty into the cost function of the LLMS to obtain a shrinkage in the weight update, which then enhances the performance in sparse system identification settings. The simulation results show that the proposed algorithm improves the performance of the filter in sparse system settings in the presence of noisy input signals.

eess.SY↗

arXiv2014

Multilinear Principal Component Analysis Network for Tensor Object Classification

The recently proposed principal component analysis network (PCANet) has been proved high performance for visual content classification. In this letter, we develop a tensorial extension of PCANet, namely, multilinear principal analysis component network (MPCANet), for tensor object classification. Compared to PCANet, the proposed MPCANet uses the spatial structure and the relationship between each dimension of tensor objects much more efficiently. Experiments were conducted on different visual content datasets including UCF sports action video sequences database and UCF11 database. The experimental results have revealed that the proposed MPCANet achieves higher classification accuracy than PCANet for tensor object classification.

cs.CV↗

arXiv2014

Tensor object classification via multilinear discriminant analysis network

This paper proposes a multilinear discriminant analysis network (MLDANet) for the recognition of multidimensional objects, known as tensor objects. The MLDANet is a variation of linear discriminant analysis network (LDANet) and principal component analysis network (PCANet), both of which are the recently proposed deep learning algorithms. The MLDANet consists of three parts: 1) The encoder learned by MLDA from tensor data. 2) Features maps ob-tained from decoder. 3) The use of binary hashing and histogram for feature pooling. A learning algorithm for MLDANet is described. Evaluations on UCF11 database indicate that the proposed MLDANet outperforms the PCANet, LDANet, MPCA + LDA, and MLDA in terms of classification for tensor objects.

cs.CV↗