Search

With the development of image recovery models,especially those based on adversarial and perceptual losses,the detailed texture portions of images are being recovered more naturally.However,these restored images are similar but not identical in detail texture to their reference images.With traditional image quality assessment methods,results with better subjective perceived quality often score lower in objective scoring.Assessment methods suffer from subjective and objective inconsistencies.This paper proposes a regional differential information entropy (RDIE) method for image quality assessment to address this problem.This approach allows better assessment of similar but not identical textural details and achieves good agreement with perceived quality.Neural networks are used to reshape the process of calculating information entropy,improving the speed and efficiency of the operation. Experiments conducted with this study image quality assessment dataset and the PIPAL dataset show that the proposed RDIE method yields a high degree of agreement with people average opinion scores compared to other image quality assessment metrics,proving that RDIE can better quantify the perceived quality of images., Comment: 14 pages, 8 figures, 5 tables

For any positive integer $t$, a \emph{$t$-broom} is a graph obtained from $K_{1,t+1}$ by subdividing an edge once. In this paper, we show that, for graphs $G$ without induced $t$-brooms, we have $\chi(G) = o(\omega(G)^{t+1})$, where $\chi(G)$ and $\omega(G)$ are the chromatic number and clique number of $G$, respectively. When $t=2$, this answers a question of Schiermeyer and Randerath. Moreover, for $t=2$, we strengthen the bound on $\chi(G)$ to $7\omega(G)^2$, confirming a conjecture of Sivaraman. For $t\geq 3$ and \{$t$-broom, $K_{t,t}$\}-free graphs, we improve the bound to $o(\omega^{t})$., Comment: 14 pages, 1 figure

Hakimi, Schmeichel, and Thomassen in 1979 conjectured that every $4$-connected planar triangulation $G$ on $n$ vertices has at least $2(n-2)(n-4)$ Hamiltonian cycles, with equality if and only if $G$ is a double wheel. In this paper, we show that every $4$-connected planar triangulation on $n$ vertices has $\Omega(n^2)$ Hamiltonian cycles. Moreover, we show that if $G$ is a $4$-connected planar triangulation on $n$ vertices and the distance between any two vertices of degree $4$ in $G$ is at least $3$, then $G$ has $2^{\Omega(n^{1/4})}$ Hamiltonian cycles., Comment: arXiv admin note: text overlap with arXiv:2104.04898

An edge-colored graph $G$ is called \textit{rainbow} if every edge of $G$ receives a different color. Given any host graph $G$, the \textit{anti-Ramsey} number of $t$ edge-disjoint rainbow spanning trees in $G$, denoted by $r(G,t)$, is defined as the maximum number of colors in an edge-coloring of $G$ containing no $t$ edge-disjoint rainbow spanning trees. For any vertex partition $P$, let $E(P,G)$ be the set of non-crossing edges in $G$ with respect to $P$. In this paper, we determine $r(G,t)$ for all host graphs $G$: $r(G,t)=|E(G)|$ if there exists a partition $P_0$ with $|E(G)|-|E(P_0,G)|

Speech enhancement at extremely low signal-to-noise ratio (SNR) condition is a very challenging problem and rarely investigated in previous works. This paper proposes a robust speech enhancement approach (UNetGAN) based on U-Net and generative adversarial learning to deal with this problem. This approach consists of a generator network and a discriminator network, which operate directly in the time domain. The generator network adopts a U-Net like structure and employs dilated convolution in the bottleneck of it. We evaluate the performance of the UNetGAN at low SNR conditions (up to -20dB) on the public benchmark. The result demonstrates that it significantly improves the speech quality and substantially outperforms the representative deep learning models, including SEGAN, cGAN fo SE, Bidirectional LSTM using phase-sensitive spectrum approximation cost function (PSA-BLSTM) and Wave-U-Net regarding Short-Time Objective Intelligibility (STOI) and Perceptual evaluation of speech quality (PESQ)., Comment: Published in Interspeech 2019

In this paper, we study the maximum spectral radius of outerplanar $3$-uniform hypergraphs. Given a hypergraph $\mathcal{H}$, the shadow of $\mathcal{H}$ is a graph $G$ with $V(G)= V(\mathcal{H})$ and $E(G) = \{uv: uv \in h \textrm{ for some } h\in E(\mathcal{H})\}$. A graph is \textit{outerplanar} if it can be embedded in the plane such that all its vertices lie on the outer face. A $3$-uniform hypergraph $\mathcal{H}$ is called \textit{outerplanar} if its shadow has an outerplanar embedding such that every hyperedge of $\mathcal{H}$ is the vertex set of an interior triangular face of the shadow. Cvetkovi\'c and Rowlinson conjectured in 1990 that among all outerplanar graphs on $n$ vertices, the graph $K_1+ P_{n-1}$ attains the maximum spectral radius. We show a hypergraph analogue of the Cvetkovi\'c-Rowlinson conjecture. In particular, we show that for sufficiently large $n$, the $n$-vertex outerplanar $3$-uniform hypergraph of maximum spectral radius is the unique $3$-uniform hypergraph whose shadow is $K_1 + P_{n-1}$., Comment: 13 pages, 3 figures

We show that if a planar graph $G$ with minimum degree at least $3$ has positive Lin-Lu-Yau Ricci curvature on every edge, then $\Delta(G)\leq 17$, which then implies that $G$ is finite. This is an analogue of a result of DeVos and Mohar [{\em Trans. Amer. Math. Soc., 2007}] on the size of planar graphs with positive combinatorial curvature., Comment: 10 pages, 2 figures

We study the existence of solutions of Ricci flow equations of Ollivier-Lin-Lu-Yau curvature defined on weighted graphs. Our work is motivated by\cite{NLLG} in which the discrete time Ricci flow algorithm has been applied successfully as a discrete geometric approach in detecting complex networks. Our main result is the existence and uniqueness theorem for solutions to a continuous time normalized Ricci flow. We also display possible solutions to the Ricci flow on path graph and prove the Ricci flow on finite star graph with at least three leaves converges to constant-weighted star., Comment: 20 pages

It is very challenging for speech enhancement methods to achieves robust performance under both high signal-to-noise ratio (SNR) and low SNR simultaneously. In this paper, we propose a method that integrates an SNR-based teachers-student technique and time-domain U-Net to deal with this problem. Specifically, this method consists of multiple teacher models and a student model. We first train the teacher models under multiple small-range SNRs that do not coincide with each other so that they can perform speech enhancement well within the specific SNR range. Then, we choose different teacher models to supervise the training of the student model according to the SNR of the training data. Eventually, the student model can perform speech enhancement under both high SNR and low SNR. To evaluate the proposed method, we constructed a dataset with an SNR ranging from -20dB to 20dB based on the public dataset. We experimentally analyzed the effectiveness of the SNR-based teachers-student technique and compared the proposed method with several state-of-the-art methods., Comment: Published in 2020 IEEE International Conference on Multimedia and Expo (ICME 2020)

3D moving object detection is one of the most critical tasks in dynamic scene analysis. In this paper, we propose a novel Drosophila-inspired 3D moving object detection method using Lidar sensors. According to the theory of elementary motion detector, we have developed a motion detector based on the shallow visual neural pathway of Drosophila. This detector is sensitive to the movement of objects and can well suppress background noise. Designing neural circuits with different connection modes, the approach searches for motion areas in a coarse-to-fine fashion and extracts point clouds of each motion area to form moving object proposals. An improved 3D object detection network is then used to estimate the point clouds of each proposal and efficiently generates the 3D bounding boxes and the object categories. We evaluate the proposed approach on the widely-used KITTI benchmark, and state-of-the-art performance was obtained by using the proposed approach on the task of motion detection.

In 1964, Erd\H{o}s, Hajnal and Moon introduced a saturation version of Tur\'an's classical theorem in extremal graph theory. In particular, they determined the minimum number of edges in a $K_r$-free, $n$-vertex graph with the property that the addition of any further edge yields a copy of $K_r$. We consider analogues of this problem in other settings. We prove a saturation version of the Erd\H{o}s-Szekeres theorem about monotone subsequences and saturation versions of some Ramsey-type theorems on graphs and Dilworth-type theorems on posets. We also consider semisaturation problems, wherein we allow the family to have the forbidden configuration, but insist that any addition to the family yields a new copy of the forbidden configuration. In this setting, we prove a semisaturation version of the Erd\H{o}s-Szekeres theorem on convex $k$-gons, as well as multiple semisaturation theorems for sequences and posets.