arXiv.itsnick.co

an independent field guide to 23 million papers from arXiv & OpenAlex — not affiliated with arXiv.org

Author

Runyi Yu

48 papers · 2003–2026

2003: 12004: 32005: 12006: 12007: 32008: 32009: 22010: 02011: 12012: 12013: 12014: 12015: 12016: 02017: 12018: 32019: 22020: 12021: 12022: 52023: 32024: 42025: 42026: 5
2003peak 5 in 20222026

Frequent co-authors

Papers

showing 1–25 of 48 RSS

Hilbert transform pairs of orthogonal wavelet bases: necessary and sufficient conditions

Runyi Yu, Hüseyin Özkaramanlı

Computer Vision and Pattern Recognition · IEEE Transactions on Signal Processing · · cited 59× · DOI ↗ · more like this ·

The condition on scaling filters of two orthogonal wavelet bases that render the corresponding wavelets as Hilbert transform pairs is re-examined in this note. Without making any pre-assumption on the relationship between the two scaling filters, the authors derive necessary and sufficient conditions for forming Hilbert transform pairs. They lead to new magnitude conditions and Selesnick's phase condition. Unique solutions to these conditions are concluded. It is shown that orthogonal wavelet bases form Hilbert transform pairs if and only if the two scaling filters are offset from one another by half a sample.

ACSeg: Adaptive Conceptualization for Unsupervised Semantic Segmentation

Kehan Li, Zhennan Wang, Zesen Cheng +6 more

cs.CV · · cited 52× · arXiv ↗ · more like this ·

Recently, self-supervised large-scale visual pre-training models have shown great promise in representing pixel-level semantic relationships, significantly promoting the development of unsupervised dense prediction tasks, e.g., unsupervised semantic segmentation (USS). The extracted relationship among pixel-level representations typically contains rich class-aware information that semantically identical pixel embeddings in the representation space gather together to form sophisticated concepts. However, leveraging the learned models to ascertain semantically consistent pixel groups or regions in the image is non-trivial since over/ under-clustering overwhelms the conceptualization procedure under various semantic distributions of different images. In this work, we investigate the pixel-level semantic aggregation in self-supervised ViT pre-trained models as image Segmentation and propose…

On the phase condition and its solution for hilbert transform pairs of wavelet bases

Hüseyin Özkaramanlı, Runyi Yu

Computer Vision and Pattern Recognition · IEEE Transactions on Signal Processing · · cited 40× · DOI ↗ · more like this ·

In this paper, the phase condition on the scaling filters of two wavelet bases that renders the corresponding wavelets as Hilbert transform pairs is studied. An alternative and equivalent phase condition is derived. With the equivalent condition and using Fourier series expansions, we show that the solution for which the corresponding scaling filters are offset from one another by a half sample is the only solution satisfying the phase condition.

Hilbert transform pairs of biorthogonal wavelet bases

Runyi Yu, Hüseyin Özkaramanlı

Computer Vision and Pattern Recognition · IEEE Transactions on Signal Processing · · cited 28× · DOI ↗ · more like this ·

The forming of Hilbert transform pairs of biorthogonal wavelet bases of two-band filter banks is studied in this paper. We first derive necessary and sufficient conditions on the scaling filters that render two Hilbert transform pairs: one decomposition pair and one reconstruction pair. We show that the Hilbert transform pairs are achieved if and only if the decomposition scaling filter of one filter bank is half-sample delayed from that of the other filter bank; and the reconstruction scaling filter of the former is half-sample advanced from that of the latter. Hilbert transform pairs of wavelet bases are also characterized by equivalent relationships on the wavelet filters and the scaling functions associated with the two filter banks. An illustrative example is provided.

A New Shift-Invariance of Discrete-Time Systems and Its Application to Discrete Wavelet Transform Analysis

Runyi Yu

Computer Vision and Pattern Recognition · IEEE Transactions on Signal Processing · · cited 22× · DOI ↗ · more like this ·

This work is motivated by the search for discrete wavelet transform (DWT) with near shift-invariance. After examining the elements of the property, we introduce a new notion of shift-invariance, which is particularly informative for multirate systems that are not shift-invariant in the strict sense. Briefly speaking, a discrete-time system is mu-shift-invariant if a shift in input results in the output being shifted as well. However, the amount of the shift in output is not necessarily identical to that in input. A fractional shift is also acceptable and can be properly specified in the Fourier domain. The mu-shift-invariance can be interpreted as invariance of magnitude spectrum with linear phase offset of output with respect to shift in input. It is stronger than the shiftability in position, which is equivalent to insensitivity of energy to shift in input. Under this generalized…

Theory of Dual-Tree Complex Wavelets

Runyi Yu

Computer Vision and Pattern Recognition · IEEE Transactions on Signal Processing · · cited 22× · DOI ↗ · more like this ·

We study analyticity of the complex wavelets in Kingsbury's dual-tree wavelet transform. A notion of scaling transformation function that defines the relationship between the primal and dual scaling functions is introduced and studied in detail. The analyticity property is examined and dealt with via the transformation function. We separate analyticity from other properties of the wavelet such as orthogonality or biorthogonality. This separation allows a unified treatment of analyticity for general setting of the wavelet system, which can be dyadic or M-band; orthogonal or biorthogonal; scalar or multiple; bases or frames. We show that analyticity of the complex wavelets can be characterized by scaling filter relationship and wavelet filter relationship via the scaling transformation function. For general orthonormal wavelets and dyadic biorthogonal scalar wavelets, the transformation…

Unlimited-Size Diffusion Restoration

Yinhuai Wang, Jiwen Yu, Runyi Yu +1 more

cs.CV · · cited 18× · arXiv ↗ · more like this ·

Recently, using diffusion models for zero-shot image restoration (IR) has become a new hot paradigm. This type of method only needs to use the pre-trained off-the-shelf diffusion models, without any finetuning, and can directly handle various IR tasks. The upper limit of the restoration performance depends on the pre-trained diffusion models, which are in rapid evolution. However, current methods only discuss how to deal with fixed-size images, but dealing with images of arbitrary sizes is very important for practical applications. This paper focuses on how to use those diffusion-based zero-shot IR methods to deal with any size while maintaining the excellent characteristics of zero-shot. A simple way to solve arbitrary size is to divide it into fixed-size patches and solve each patch independently. But this may yield significant artifacts since it neither considers the global semantics…

Residual Correlation Regularization Based Image Denoising

Gulsher Baloch, Hüseyin Özkaramanlı, Runyi Yu

Computer Vision and Pattern Recognition · IEEE Signal Processing Letters · · cited 16× · DOI ↗ · more like this ·

Patch-based denoising algorithms aim to reconstruct the clean image patch leaving behind the residual as contaminating noise. The residual should possess statistical properties of contaminating noise. However, it is very likely that the residual patch contains remnants from the clean image patch. In this letter, we propose a new residual correlation based regularization for image denoising. The regularization can effectively render residual patches as uncorrelated as possible. It allows us to derive an analytical solution for sparse coding (atom selection and coefficient calculation). It also leads to a new online dictionary learning update. The clean image is obtained through alternating between the two stages of sparse coding and dictionary updating. The performance of the proposed algorithm is compared with state-of-the-art denoising algorithms in terms of peak signal-to-noise ratio…

Sampled-Data Design of FIR Dual Filter Banks for Dual-Tree Complex Wavelet Transforms via LMI Optimization

Runyi Yu, Aryaz Baradarani

Computer Vision and Pattern Recognition · IEEE Transactions on Signal Processing · · cited 14× · DOI ↗ · more like this ·

Starting from a given finite-impulse-response (FIR) primal filter bank, we design a dual filter bank such that the complex wavelets associated with the dual-tree filter bank are (almost) analytic. The dual filter bank is required to be FIR and have a prescribed number of zeros at . We formulate a sampled-data optimization problem based on the half-sample delay condition on scaling filters. A discrete-time filter is introduced in the formulation to specify the number of the zeros. The optimization problem is converted into an equivalent discrete-time control problem; the latter is further reduced to an LMI optimization problem. We then present a procedure for design of FIR dual filter banks. Illustrative examples are provided; the results compare favorably to early designs.

Characterization and Sampled-Data Design of Dual-Tree Filter Banks for Hilbert Transform Pairs of Wavelet Bases

Runyi Yu

Computer Vision and Pattern Recognition · IEEE Transactions on Signal Processing · · cited 13× · DOI ↗ · more like this ·

Characterization and design of dual-tree filter banks for forming Hilbert transform pairs of wavelet bases are studied. The characterization extends the existing results for quadrature mirror filter banks to general prefect reconstruction filter banks that satisfy only a mild technical assumption regarding the ratio of determinants of the two filter banks. We establish equivalent relationships of Hilbert transform pairs on scaling filters, wavelet filters, or scaling functions. The design of scaling filters of a dual filter bank is formulated as a sampled-data H <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">infin</sup> optimization problem. The wavelet filters are then determined using the relationship on the determinants of the filter banks. We convert the sampled-data problem into an equivalent discrete-time H <sup…

Design of Halfband Filters for Orthogonal Wavelets via Sum of Squares Decomposition

Runyi Yu, Aryaz Baradarani

Computer Vision and Pattern Recognition · IEEE Signal Processing Letters · · cited 12× · DOI ↗ · more like this ·

This letter is concerned with design of halfband product filters for orthogonal wavelets. We first remark that the recent zero-pinning technique for orthogonal wavelet design cannot always guarantee the nonnegativity of the filter. We then propose to use sum of squares (SOS) decomposition to ensure its nonnegativity. The use of SOS decomposition also allows us to solve two optimization problems on the halfband filter via semidefinite programming. For a given length with pre-specified number of zeros at , we obtain halfband filters with maximal passband width. Design examples are provided.

A Dual-Tree Complex Wavelet with Application in Image Denoising

Aryaz Baradarani, Runyi Yu

Computer Vision and Pattern Recognition · · cited 12× · DOI ↗ · more like this ·

This paper introduces a recently designed dual-tree complex wavelet and studies its application in image denoising. The primal filter bank is selected to be the Daubechies 9/7 filter bank, and the dual filter bank is designed to have length of 10/8; both filter banks are biorthogonal and symmetric. The wavelets of the dual-tree filter bank form (almost) Hilbert transform pairs, allowing nearly shift-invariance and good directionality of the dual-tree complex wavelet transform. The transform is then used in image denoising. We employ the bivariate shrinkage algorithm for wavelet coefficient modeling and thresholding. Various images are tested. The experimental results compare favorably to some other dual-tree complex wavelets.

On the phase condition and its solution for Hilbert transform pairs of wavelets bases

Hüseyin Özkaramanlı, Runyi Yu

Computer Vision and Pattern Recognition · 2003 IEEE International Conference on Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03). · · cited 12× · DOI ↗ · more like this ·

In this work, the phase condition on the scaling filters of two wavelet bases which ensures that the corresponding wavelets are Hilbert transform pairs of each other is studied. An alternative and equivalent phase condition is derived. With the equivalent condition and Fourier series expansions, it is shown that the solution for which the corresponding scaling filters are offset from one another by a half sample is the only solution satisfying the phase condition which results in Hilbert transform pairs of wavelet bases.

Neural Network-Based Video Compression Artifact Reduction Using Temporal Correlation and Sparsity Prior Predictions

Wei‐Gang Chen, Runyi Yu, Xun Wang

Computer Vision and Pattern Recognition · IEEE Access · · cited 10× · DOI ↗ · more like this ·

Quantization in lossy video compression may incur severe quality degradation, especially at low bit-rates. Developing post-processing methods that improve visual quality of decoded images is of great importance, as they can be directly incorporated in any existing compression standard or paradigm. We propose in this article a two-stage method, a texture detail restoration stage followed by a deep convolutional neural network (CNN) fusion stage, for video compression artifact reduction. The first stage performs in a patch-by-patch manner. For each patch in the current decoded frame, one prediction is formed based on the sparsity prior assuming that natural image patches can be represented by sparse activation of dictionary atoms. Under the temporal correlation hypothesis, we search the best matching patch in each reference frame, and select several matches with more texture details to…

Shift-Variance Analysis of Generalized Sampling Processes

Runyi Yu

Computer Vision and Pattern Recognition · IEEE Transactions on Signal Processing · · cited 8× · DOI ↗ · more like this ·

This paper is concerned with quantifying shift-variance of linear systems with continuous-time input and discrete-time output. We first introduce a notion of -shift-invariance for the system. It specifies how the system should respond when the input signal is shifted. For generalized sampling processes, the property is characterized by the sampling kernel, and is shown to be equivalent to lack of aliasing and to shiftability. We then define a shift-variance level which describes how far the system can be possibly away from the set of systems that are -shift-invariant. The level is defined to be the maximum of the induced norms of the commutators of the system and the shift-operators (both continuous-time or discrete-time are necessarily involved). A shift-variance measure is then defined to be the ratio of the shift-variance level to the system norm, further divided by two so that the…

LaPE: Layer-adaptive Position Embedding for Vision Transformers with Independent Layer Normalization

Runyi Yu, Zhennan Wang, Yinhuai Wang +5 more

Computer Vision and Pattern Recognition · · cited 7× · DOI ↗ · more like this ·

Position information is critical for Vision Transformers (VTs) due to the permutation-invariance of self-attention operations. A typical way to introduce position information is adding the absolute Position Embedding (PE) to patch embedding before entering VTs. However, this approach operates the same Layer Normalization (LN) to token embedding and PE, and delivers the same PE to each layer. This results in restricted and monotonic PE across layers, as the shared LN affine parameters are not dedicated to PE, and the PE cannot be adjusted on a per-layer basis. To overcome these limitations, we propose using two independent LNs for token embeddings and PE in each layer, and progressively delivering PE across layers. By implementing this approach, VTs will receive layer-adaptive and hierarchical PE. We name our method as Layer-adaptive Position Embedding, abbreviated as LaPE, which is…

A Tensor Approach to Model Order Selection of Multiple Sinusoids

Erhan A. İnce, Mehrab K. Allahdad, Runyi Yu

Signal Processing · IEEE Signal Processing Letters · · cited 7× · DOI ↗ · more like this ·

The letter presents a covariance tensor based estimator that can effectively provide the model order for multiple sinusoids in noise. For moderate to high signal-to-noise ratios the estimator works effectively even for short length observations. Unlike the conventional methods that use a covariance matrix, the proposed estimator makes use of a three-way covariance tensor formed from the one-dimensional data. Estimate of the model order has been determined by processing the incremental Frobenius norms of the leading principal subtensors of the core tensor obtained via higher order singular value decomposition. Simulations under additive white Gaussian noise show that the new estimator offers significant improvements over conventional methods such as multiple signal classification (MUSIC), estimation of signal parameters via rotational invariance technique+maximum a posteriori, and…

Shifting Interpolation Kernel Toward Orthogonal Projection

Bashir Sadeghi, Runyi Yu, Ruili Wang

Computer Vision and Pattern Recognition · IEEE Transactions on Signal Processing · · cited 7× · DOI ↗ · more like this ·

Orthogonal projection offers the optimal solution for many sampling-reconstruction problems in terms of the least square error. In the standard interpolation setting where the sampling is assumed to be ideal, however, the projection is impossible unless the interpolation kernel is related to the sinc function and the input is bandlimited. In this paper, we propose a notion of shifting kernel toward the orthogonal projection. For a given interpolation kernel, we formulate optimization problems whose solutions lead to shifted interpolations that, while still being interpolatory, are closest to the orthogonal projection in the sense of the minimax regret. The quality of interpolation is evaluated in terms of the average approximation error over input shift. For the standard linear interpolation, we obtain several values of optimal shift, dependent on a priori information on input signals.…

Shift-Variance and Nonstationarity of Linear Periodically Shift-Variant Systems and Applications to Generalized Sampling-Reconstruction Processes

Bashir Sadeghi, Runyi Yu

Computer Vision and Pattern Recognition · IEEE Transactions on Signal Processing · · cited 7× · DOI ↗ · more like this ·

We present a systematic analysis on shift-variance of linear T-periodically shift-variant ( T-LPSV) systems with continuous-time input and output. We first determine how far a T-LPSV system H is away from being shift-invariant in terms of two measures, namely, the distance and the angle between H and the subspace of linear shift-invariant systems. We then consider the level and the index of shift-variance (SVI) of H via commutator of H and the shift-operator, and examine the amount of shift-variance of H under particular input. All these quantities are characterized in terms of the so-called shift-variant part and the shift-variance kernel of H. Moreover, we study nonstationarity of wide-sense cyclostationary random processes, possibly induced by T-LPSV systems. We define and calculate a measure of nonstationarity (NSt) of the output as the SVI of the autocorrelation operator. The Nst…

Shift-Variance Measure of Multichannel Multirate Systems

Runyi Yu

Computer Vision and Pattern Recognition · IEEE Transactions on Signal Processing · · cited 6× · DOI ↗ · more like this ·

This study is concerned with quantifying shift variance of linear discrete-time multirate systems. Based on commutators of a system and the shift operators, we introduce a shift-variance measure for the multirate system. The measure is defined as the half of the ratio of the maximal norm of the commutators to the norm of the multirate system. It ranges from zero to unity, corresponding to minimal shift variance (i.e., full shift invariance) and maximal shift variance, respectively. We derive analytical expressions on the shift-variance measure of multirate filterbanks. The relation between shift variance of the individual channels of multichannel filterbanks is then studied. We also analyze the shift invariance of some particular filterbanks, including biorthogonal filterbanks and complex wavelet transforms.

Unified Motion Retargeting for Humanoids with Learned Point Cloud Correspondence

Hanyang Cao, Yuetong Fang, Taesoo Kwon +10 more

cs.RO · · cited 5× · arXiv ↗ · more like this ·

Humanoid learning increasingly relies on transforming vast and diverse human motion data into high-quality robot reference trajectories. However, retargeting human motion to humanoid robots is challenging due to substantial differences in morphology, degrees of freedom, joint ranges, and kinematic constraints between humans and robots. Existing retargeting methods typically address these differences by defining human-robot correspondence through hand-crafted sparse keypoints or body-part pairs. As a result, retargeting quality depends heavily on manual semantic design, limiting scalability across motion sources and robot morphologies and providing only sparse guidance for reproducing detailed poses and interactions. In this paper, we present Unified Motion Retargeting (UMR), a framework that learns dense point cloud correspondence without requiring manually designed human-robot…

Subspace Alignment and Separation for Multiple Frequency Estimation

Runyi Yu, Erhan A. İnce, Aykut Hocanın

Signal Processing · IEEE Signal Processing Letters · · cited 4× · DOI ↗ · more like this ·

In this letter, a new subspace based estimator that can effectively provide the order and frequencies of multiple sinusoids in noise is proposed. The estimator, referred to as SAS-Est (Subspace Aligning and Separating Estimator), simultaneously seeks to separate the steering vectors from the noise subspace and align them to the signal subspace. The angles between subspaces and the generalized Kullback-Leibler divergence are used in characterizing the alignment and separation. Minimizing the divergence leads to maximal subspace separation and best alignment, thus allowing improved performance. Simulations in additive white Gaussian noise show that the new estimator offers an improvement for both model order and frequency estimation. When compared with other methods, the improvement is more pronounced for high model orders and low signal-to-noise ratio values.

Image inpainting via singular value thresholding

S. Faegheh Yeganli, Runyi Yu

Computer Vision and Pattern Recognition · · cited 4× · DOI ↗ · more like this ·

This paper proposes an inpainting technique based on the singular value thresholding (SVT) algorithm for matrix completion. The SVT algorithm computes the missing elements of a matrix so that the resultant matrix is an approximate of the completion with least rank. It can be implemented by a straightforward iteration procedure which performs singular value decompositions followed by a soft thresholding on the singular values. Experimental results show that the technique can effectively recover texture parts of damaged images.

HiPHI: A Large-Scale Benchmark for High-Precision Human Motion and Object-Interaction

Jiahao Ji, Ji Ma, Runhan Zhang +12 more

cs.RO · · cited 2× · arXiv ↗ · more like this ·

Humanoid intelligence requires learning over an extremely diverse space of whole-body motions and physically grounded interactions. However, existing embodied datasets remain fundamentally limited: internet-scale video data lack precise physical states and interaction grounding, while laboratory motion datasets provide high fidelity but only narrow behavioral coverage. This mismatch creates a critical bottleneck for scalable humanoid policy learning. We present HiPHI, a 600+ hour scale high-fidelity whole-body human motion dataset designed to systematically maximize coverage of the human motion and interaction manifold. HiPHI is theoretically guided by FrameNet, a linguistic framework organizing human primitives. Created using an optical motion capture pipeline, HiPHI provides sub-millimeter spatial marker tracking accuracy for full-body human motion and mesh-level object trajectories.…

A Reduced Complexity Tensor Approach for Order Selection and Frequency Estimation

Runyi Yu, Erhan A. İnce

Signal Processing · IEEE Access · · cited 2× · DOI ↗ · more like this ·

This paper presents a reduced complexity tensor approach for order selection and subspace-based frequency estimation. The proposed covariance tensor based order selection algorithm, termed as CT-OS, uses singular values of the covariance tensor formed from the 1D noisy observations of multiple complex sinusoids. Experimental results show that the CT-OS algorithm is capable of providing accurate order selection under short observations and is robust under medium to high signal-to-noise ratios. The proposed covariance tensor based frequency estimator, termed as CT-FE, utilizes a singular vector matrix in the higher-order singular value decomposition of the covariance tensor. Experimental results show that the CT-FE outperforms the subspace alignment and separation algorithm (SAS-Est) and a recent two-stage order and frequency estimation algorithm. Furthermore, both theoretical analysis…