Kai Gao

ORCID: 0000-0002-9605-4964
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Advanced Vision and Imaging
  • Welding Techniques and Residual Stresses
  • Image Enhancement Techniques
  • Advanced Measurement and Detection Methods
  • Advanced Neural Network Applications
  • Non-Destructive Testing Techniques
  • CCD and CMOS Imaging Sensors
  • Advanced machining processes and optimization
  • AI and Big Data Applications
  • Solar Radiation and Photovoltaics
  • Sentiment Analysis and Opinion Mining
  • Advanced Image Processing Techniques
  • Photovoltaic System Optimization Techniques
  • Scientific Computing and Data Management
  • Cell Image Analysis Techniques
  • Ocular and Laser Science Research
  • Semiconductor Lasers and Optical Devices
  • Advanced Optical Imaging Technologies
  • Industrial Vision Systems and Defect Detection
  • Optical Coherence Tomography Applications
  • Open Source Software Innovations
  • Optical measurement and interference techniques
  • Digital Media and Visual Art
  • Digital Holography and Microscopy
  • Image Processing Techniques and Applications

Hebei University of Science and Technology
2024

Institute of Basic Medical Sciences of the Chinese Academy of Medical Sciences
2024

Changchun University of Science and Technology
2015-2024

Beijing Institute of Education
2024

Shandong University of Art and Design
2022

Jiangsu University of Science and Technology
2014-2015

Jilin University
2010-2013

Jilin Medical University
2010

10.1007/s00170-014-6678-6 article EN The International Journal of Advanced Manufacturing Technology 2014-12-09

In multimodal sentiment analysis, achieving effective fusion among text, acoustic, and visual modalities for enhanced prediction is a crucial research topic. Recent studies typically employ tensor-based or attention-based mechanisms fusion. However, the former fails to achieve satisfactory performance, latter complicates computation of between non-textual modalities. Therefore, this paper proposes analysis model based on Adaptive Feature Normalization Attention Gating mechanism (AdaFN-AG)....

10.2139/ssrn.4821987 preprint EN 2024-01-01

For the purpose of representing feature gray image, illumination normal pixels in a two-dimensional image plane is proposed, which can reflect high-frequency information image. In order to get an accurate dense disparity map based on adaptive support weight (ASW) approach RGB vector space, matching algorithm proposed that combines similarity, gradient color and Euclidean distance similarity compute corresponding weights dissimilarity measurements. After testing by Middlebury stereo...

10.1117/1.oe.52.2.027201 article EN cc-by Optical Engineering 2013-02-01

Abstract In practical operating environments, it is extremely important for mobile robots to reach the target area quickly and safely. Therefore, a new path planning algorithm, APF-EB-RRT, based on EB-RRT algorithm was proposed in this article. The artificial potential field introduced into APF-EB-RRT algorithm. By adding constraints that simulate interaction between objects dynamic reprogramming structure, can avoid problems of high randomness poor directionality caused by EB Meanwhile,...

10.21203/rs.3.rs-3872824/v1 preprint EN cc-by Research Square (Research Square) 2024-01-19

This paper proposes a simplified method to generate elemental images using the image pair of scene for integral imaging stereo display. Firstly, generalized superpixel matching algorithm is designed obtain depth map. We define discrete boundary according characteristics low-texture region. A kind consistency check by calculating inclination degree proposed. The remaining texture regions are handled assisted growth method. precise map obtained through above two steps. Secondly, and color used...

10.35848/1347-4065/ac2a00 article EN Japanese Journal of Applied Physics 2021-09-24

Depth map is critical in Free-viewpoint television (FTV) system, and the quality of reconstructed depth impacts rendering view. obtained from TOF camera, not only appears with large flat area sharp edges, but also contains lots noises. In order to achieve aim decreasing noise keeping accurate edges map, an iterative trilateral filter proposed by combining bilateral introduced factor illumination normal this paper. The experimental results show that method can reduce obviously,and keep edge...

10.1117/12.2175482 article EN Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE 2015-03-04

The reuse and distribution of open-source software must be in compliance with its accompanying license. In modern packaging ecosystems, maintaining such is challenging because a package may have complex multi-layered dependency graph many packages, any which an incompatible Although prior research finds that license incompatibilities are prevalent, empirical evidence still scarce some ecosystems (e.g., PyPI). It also remains unclear how developers remediate the graphs their packages...

10.48550/arxiv.2308.05942 preprint EN cc-by-nc-sa arXiv (Cornell University) 2023-01-01

The bit error ratio (BER) and power are very hard to calculate in space optical communications, having lots of effect factors, so need compromise consider the complex factors. Analysis influence BER such as beam drift, divergence angle, communication distance, link loss, detector sensitivity etc. Propose view ellipse gauss can inhibit drift by random shock relative motion platform reduce BER, further propose method transmit using BER. Experiments showed that it is security use calculation...

10.1117/12.868266 article EN Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE 2010-11-03

Due to the large amount of noise in image background, traditional methods have problems poor communication effect, slow efficiency, and processing effect. A three-dimensional method art visual based on CAD platform is proposed. Wavelet transform used extract artistic features, fuzzy clustering standardize obtained feature data. Based data results, constructed, software program designed, chain coding realized at same time. The particle swarm optimization algorithm convey general vectorization...

10.1155/2022/3441252 article EN Scientific Programming 2022-02-27

In order to achieve high precision stereo matching in the case of limited computing resources, an end-to-end parallax network is proposed by improving DispNet network. The uses code-decoding structure, adds residual structure and cost calculation method based on matrix multiplication optimize ECA module estimation for training, adaptively enhances important channel information body, improves learning ability experimental results show that improved can dense high-precision prediction real...

10.1109/icftic54370.2021.9647340 article EN 2021 IEEE 3rd International Conference on Frontiers Technology of Information and Computer (ICFTIC) 2021-11-12
Coming Soon ...