A5100434894- Advanced Vision and Imaging
- Advanced Image and Video Retrieval Techniques
- Advanced Image Processing Techniques
- Optical measurement and interference techniques
- Image Enhancement Techniques
- Robotics and Sensor-Based Localization
- Image Processing Techniques and Applications
- Image Retrieval and Classification Techniques
- Computer Graphics and Visualization Techniques
- Video Analysis and Summarization
- Video Surveillance and Tracking Methods
- Human Pose and Action Recognition
- Image and Signal Denoising Methods
- Video Coding and Compression Technologies
- Handwritten Text Recognition Techniques
- Remote Sensing and LiDAR Applications
- Generative Adversarial Networks and Image Synthesis
- Anomaly Detection Techniques and Applications
- Image Processing and 3D Reconstruction
- Vehicle License Plate Recognition
- Image and Video Quality Assessment
- Speech Recognition and Synthesis
- Image and Object Detection Techniques
- Industrial Vision Systems and Defect Detection
- Face and Expression Recognition
Northwestern Polytechnical University
2016-2025
China National Petroleum Corporation (China)
2020-2024
Zhejiang Normal University
2024
Shanghai Academy of Environmental Sciences
2024
Yangtze University
2019-2023
Anhui University of Traditional Chinese Medicine
2021
China Agricultural University
2018-2020
Sun Yat-sen University
2012-2019
Qilu University of Technology
2019
Tianjin University
2018
A fast and effective method for image contrast enhancement is presented. In the proposed method, probability distribution function (histogram) of an modified by weighting thresholding before histogram equalization (HE) performed. We show that such approach provides a convenient mechanism to control process, while being adaptive various types images. also discuss application in video enhancement. Experimental results are presented compared with from other contemporary methods.
<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> A new universal estimator of divergence is presented for multidimensional continuous densities based on <emphasis emphasistype="italic"><formula formulatype="inline"> <tex Notation="TeX">$k$</tex></formula></emphasis>-nearest-neighbor (<emphasis formulatype="inline"><tex Notation="TeX">$k$</tex> </formula></emphasis>-NN) distances. Assuming independent and identically distributed (i.i.d.)...
We present High Dynamic Range Neural Radiance Fields (HDR-NeRF) to recover an HDR radiance field from a set of low dynamic range (LDR) views with different exposures. Using the HDR-NeRF, we are able generate both novel and LDR under The key our method is model simplified physical imaging process, which dictates that scene point transforms pixel value in image two implicit functions: tone mapper. encodes (values vary 0 <tex xmlns:mml="http://www.w3.org/1998/Math/MathML"...
We propose an isophote-oriented, orientation-adaptive interpolation method. The proposed method employs kernel that adapts to the local orientation of isophotes, and pixel values are obtained through oriented, bilinear interpolation. show that, by doing so, curvature interpolated isophotes is reduced, and, thus, zigzagging artifacts largely suppressed. Analysis experiments images using visually pleasing almost artifact free
This paper presents the results of depth estimation challenge for dense light fields, which took place at second workshop on Light Fields Computer Vision (LF4CV) in conjunction with CVPR 2017. The consisted submission to a recent benchmark [7], allows thorough performance analysis. While individual are readily available web page http://www.lightfield-analysis.net, we take this opportunity give detailed overview current participants. Based algorithms submitted our challenge, develop taxonomy...
Image relighting is attracting increasing interest due to its various applications. From a research perspective, im-age can be exploited conduct both image normalization for domain adaptation, and also data augmentation. It has multiple direct uses photo montage aesthetic enhancement. In this paper, we review the NTIRE 2021 depth guided challenge.We rely on VIDIT dataset each of our two challenge tracks, including information. The first track one-to-one where goal transform illumination...
This paper reviews the first NTIRE challenge on quality enhancement of compressed video, with a focus proposed methods and results.In this challenge, new Large-scale Diverse Video (LDV) dataset is employed.The has three tracks.Tracks 1 2 aim at enhancing videos by HEVC fixed QP, while Track 3 designed for x265 bit-rate.Besides, Tracks targets improving fidelity (PSNR), perceptual quality.The tracks totally attract 482 registrations.In test phase, 12 teams, 8 teams 11 submitted final results...
To deal with the great number of untrimmed videos produced every day, we propose an efficient unsupervised action segmentation method by detecting boundaries, named boundary detection (ABD). In particular, proposed has following advantages: no training stage and low-latency inference. detect estimate similarities across smoothed frames, which inherently have properties internal consistency within actions external discrepancy actions. Under this circumstance, successfully transfer task into...
Being a form of biometric identification, the security speaker identification (SID) system is utmost importance. To better understand robustness SID systems, we aim to perform more realistic attacks in SID, which are challenging for both humans and machines detect. In this study, propose DiffAttack, novel timbre-reserved adversarial attack approach that exploits capability diffusion-based voice conversion (DiffVC) model generate fake audio with distinct target attribution. By introducing...
Abstract Micro-bubbles have attracted much attention in wastewater treatment due to their strong stability, long residence time water, high mass transfer efficiency, and Zeta potential [1-2]. Among the commonly used bubble generation methods, ultrasonic method, Venturi micro-dispersed method are most popular methods [3–4]. To improve efficiency of enhance degree fineness, this study proposed a new tube generator, mainly by adding multiple throat pipes based on traditional increase liquid...
Fertilization equipment is a crucial component of irrigation systems, and its performance directly impacts quality. To address the issues low suction flow excessive head loss in existing Venturi fertilizer injectors, team employed numerical simulation techniques. Through single-factor experiments, we analyzed effect pipe angle dual-throat injector on performance, ultimately determining optimal structural parameter combination for injector. The study utilized software such as Tecplot CFD to...
Our work aims to reconstruct the appearance and geometry of two-hand-held object from a sequence color images. In contrast traditional single-hand-held manipulation, two-hand-holding allows more flexible interaction, thereby providing back views object, which is particularly convenient for reconstruction but generates complex view-dependent occlusions. The recent development neural rendering provides new potential hand-held reconstruction. this paper, we propose novel representation-based...
Light field cameras can capture both spatial and angular information of light rays, enabling 3D reconstruction by a single exposure. The geometry is affected intrinsic parameters camera significantly. In the paper, we propose multi-projection-center (MPC) model with 6 to characterize based on traditional two-parallel-plane (TPP) representation. MPC generally parameterize in different imaging formations, including conventional focused cameras. By constraints 4D ray geometry, projective...
Light field cameras (LFCs) have received increasing attention due to their wide-spread applications. However, current LFCs suffer from the well-known spatio-angular trade-off, which is considered an inherent and fundamental limit for LFC designs. In this article, by doing a detailed optical analysis of sampling process in LFC, we show that effective resolution generally higher than number micro-lenses. This contribution makes it theoretically possible super-resolve light field. Further...
Superpixel segmentation of 2D image has been widely used in many computer vision tasks. However, limited to the Gaussian imaging principle, there is not a thorough solution ambiguity defocus and occlusion boundary areas. In this paper, we consider essential element pixel, i.e., rays light space propose field superpixel (LFSP) eliminate ambiguity. The LFSP first defined mathematically then refocus-invariant metric named self-similarity proposed evaluate performance. By building clique system...
Efficient and accurate segmentation of full 4D light fields is an important task in computer vision graphics. The massive volume the redundancy make it open challenge. In this article, we propose a novel field hypergraph (LFHG) representation using super-pixel (LFSP) for interactive segmentation. LFSPs not only maintain spatio-angular consistency, but also greatly contribute to coarsening. These advantages useful improve performance. Based on LFHG representation, present efficient algorithm...
This paper focuses on exploring distinctive spatio-temporal representation in a self-supervised manner for group activity recognition. Firstly, previous networks treat spatial- and temporal-aware information as whole, limiting their abilities to represent complex correlations activity. Here, we propose the Spatial Temporal Attention Heads (STAHs) extract representations independently, which generate complementary contexts boosting understanding. Then, Global Spatio-Temporal Contrastive...
We propose LIRF (Local Implicit Ray Function), a generalizable neural rendering approach for novel view rendering. Current radiance fields (NeRF) methods sample scene with single ray per pixel and may therefore render blurred or aliased views when the input rendered capture content different resolutions. To solve this problem, we to aggregate information from conical frustums construct ray. Given 3D positions within frustums, takes coordinates features of as inputs predicts local volumetric...
Abstract Spatial transcriptomics reveals the spatial distribution of genes in complex tissues, providing crucial insights into biological processes, disease mechanisms, and drug development. The prediction gene expression based on cost-effective histology images is a promising yet challenging field research. Existing methods for from exhibit two major limitations. First, they ignore intricate relationship between cell morphological information expression. Second, these do not fully utilize...