Jun Cao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Jun Cao has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
Di Jia
School of Electronic and Information Engineering Liaoning Technical University Liaoning 125105 People's Republic of China
1 shared publication- Di Jia
Ordos Research Institute and School of Electronics and Information Engineering, Liaoning Technical University, LiaoNing, China
1 shared publication - Di Jia
Liaoning Technical University
1 shared publication Hong Zhu
School of Geomatics Liaoning Technical University Liaoning 125105 People's Republic of China
1 shared publication- Mingyuan Zhao
Liaoning Technical University
1 shared publication - Peng Cai
Liaoning Technical University
1 shared publication
Latest Publications
- Open Access
- 26 June 2023
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 37 (13) , 16455-16457
FC-TrackNet: Fast Convergence Net for 6D Pose Tracking in Synthetic Domains
In this work, we propose a fast convergence track net, or FC-TrackNet, based on a synthetic data-driven approach to maintaining long-term 6D pose tracking. Comparison experiments are performed on two different datasets, The results demonstrate that our approach can achieve a consistent tracking frequency of 90.9 Hz as well as higher accuracy than the state-of-the art approaches.
- 26 June 2023
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 37 (13) , 16455-16457
Open AccessFC-TrackNet: Fast Convergence Net for 6D Pose Tracking in Synthetic Domains
In this work, we propose a fast convergence track net, or FC-TrackNet, based on a synthetic data-driven approach to maintaining long-term 6D pose tracking. Comparison experiments are performed on two different datasets, The results demonstrate that our approach can achieve a consistent tracking frequency of 90.9 Hz as well as higher accuracy than the state-of-the art approaches.
- Open Access
- 1 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Chinese Journal of Electronics
- Vol. 28 (3) , 598-603
A Method of Non‐textured Regions Matching
It is difficult for existing dense or quasidense matching algorithms to obtain accurate matching results due to the lack of effective feature information in non-textured regions. This may affect the quality of subsequent 3D reconstruction and super-resolution reconstruction. We propose a method of non-textured regions matching. To reduce the impact of noise and illumination, vector sampling normalized cross correlation is proposed to directly measure coherence between two colour images by the effective information of multi-channel features. Three mathematical properties are used in affine transformation: 1) Centroid location is not affected by affine transformation; 2) Affine transformation transforms a straight line into another straight line; 3) Affine transformation maintains the linear relation invariant. In non-textured regions, we can construct texture regions which have affine invariant properties to improve the accuracy of template matching. This could provide more information for the diffusion of dense or quasi-dense matching. In experiments, we demonstrate that the method has high accuracy in cases where there is no big distortion between the two viewing angles from two aspects of simulated images and photographic images.
- 1 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Chinese Journal of Electronics
- Vol. 28 (3) , 598-603
Open AccessA Method of Non‐textured Regions Matching
It is difficult for existing dense or quasidense matching algorithms to obtain accurate matching results due to the lack of effective feature information in non-textured regions. This may affect the quality of subsequent 3D reconstruction and super-resolution reconstruction. We propose a method of non-textured regions matching. To reduce the impact of noise and illumination, vector sampling normalized cross correlation is proposed to directly measure coherence between two colour images by the effective information of multi-channel features. Three mathematical properties are used in affine transformation: 1) Centroid location is not affected by affine transformation; 2) Affine transformation transforms a straight line into another straight line; 3) Affine transformation maintains the linear relation invariant. In non-textured regions, we can construct texture regions which have affine invariant properties to improve the accuracy of template matching. This could provide more information for the diffusion of dense or quasi-dense matching. In experiments, we demonstrate that the method has high accuracy in cases where there is no big distortion between the two viewing angles from two aspects of simulated images and photographic images.
- Open Access
- 1 July 2016
- Published by Institution of Engineering and Technology (IET) in Electronics Letters
- Vol. 52 (14) , 1220-1221
Colour FAST (CFAST) match: fast affine template matching for colour images
Fast‐match is a fast and effective algorithm for template matching. However, when matching colour images, the images are converted into greyscale images. The colour information is lost in this process, resulting in errors in areas with distinctive colours but similar greyscale values An improved fast‐match algorithm that utilises all three RGB channels to construct colour sum‐of‐absolute‐differences (CSAD) is proposed, thus improving the sum‐of‐absolute‐differences distance used in fast‐match. In this algorithm, each pixel in the image is categorised by clustering them using density‐based spatial clustering of applications with noise (DBSCAN) algorithm over the RGB vector, then the number of pixels in each category and the cumulative RGB values for each RGB channel are calculated to identify the centroid of each category. The RGB vector centroid is used as the CSAD decision criteria, and inverse of number of pixels in each category is used as the differentiating coefficient to construct a new similarity measure. Experiment results demonstrate that this algorithm has significant higher accuracy for matching colour images than the original fast‐match algorithm.
- 1 July 2016
- Published by Institution of Engineering and Technology (IET) in Electronics Letters
- Vol. 52 (14) , 1220-1221
Open AccessColour FAST (CFAST) match: fast affine template matching for colour images
Fast‐match is a fast and effective algorithm for template matching. However, when matching colour images, the images are converted into greyscale images. The colour information is lost in this process, resulting in errors in areas with distinctive colours but similar greyscale values An improved fast‐match algorithm that utilises all three RGB channels to construct colour sum‐of‐absolute‐differences (CSAD) is proposed, thus improving the sum‐of‐absolute‐differences distance used in fast‐match. In this algorithm, each pixel in the image is categorised by clustering them using density‐based spatial clustering of applications with noise (DBSCAN) algorithm over the RGB vector, then the number of pixels in each category and the cumulative RGB values for each RGB channel are calculated to identify the centroid of each category. The RGB vector centroid is used as the CSAD decision criteria, and inverse of number of pixels in each category is used as the differentiating coefficient to construct a new similarity measure. Experiment results demonstrate that this algorithm has significant higher accuracy for matching colour images than the original fast‐match algorithm.