|
|
论文:2021,Vol:39,Issue(4):876-882 |
|
|
引用本文： |
|
|
陈星, 张文海, 候宇, 杨林. 改进的基于多尺度融合的立体匹配算法[J]. 西北工业大学学报 |
|
|
CHEN Xing, ZHANG Wenhai, HOU Yu, YANG Lin. Improved stereo matching algorithm based on multi-scale fusion[J]. Northwestern polytechnical university |
|
|
|
|
|
|
|
| 改进的基于多尺度融合的立体匹配算法 |
|
| 陈星1,2, 张文海1,2, 候宇3, 杨林3 |
|
1. 重庆文理学院 智能制造工程学院, 重庆 402160;
2. 重庆交通大学 机电与汽车工程学院, 重庆 400074;
3. 重庆长安工业(集团)有限责任公司 特种车辆研究所, 重庆 400023 |
| 摘要: |
| 针对局部立体匹配算法在弱纹理或视差不连续区域匹配精度低等问题，提出了一种结合卷积神经网络（CNN）和特征金字塔结构（FPN）的多尺度融合的立体匹配算法。在卷积神经网络的基础上应用了特征金字塔，实现立体图像的多尺度特征提取和融合，提高了图像块的匹配相似度；利用引导图滤波器（guided filtering）快速有效地完成代价聚合，在视差的选择阶段采用改进的动态规划（DP）算法获得初始视差图，对初始视差图精细化得到最后的视差图。所提算法在Middlebury数据集上提供的图像进行训练和测试，结果表明该算法得到的视差图具有较好的效果。 |
| 关键词:
立体匹配
神经网络
特征金字塔
多尺度
引导图滤波器
动态规划
|
|
| Improved stereo matching algorithm based on multi-scale fusion |
|
| CHEN Xing1,2, ZHANG Wenhai1,2, HOU Yu3, YANG Lin3 |
|
1. School of Intelligent Manufacturing, Chongqing University of Arts and Sciences, Chongqing 402160, China;
2. School of Mechatronics and Vehicle Engineering, Chongqing Jiaotong University, Chongqing, 400074, China;
3. Special Vehicle Research Institute, Chongqing Changan Industry(Group) Co., Ltd., Chongqing 400023, China |
| Abstract: |
| Aiming at the low matching accuracy of local stereo matching algorithm in weak texture or discontinuous disparity areas, a stereo matching algorithm combining multi-scale fusion of convolutional neural network (CNN) and feature pyramid structure (FPN) is proposed. The feature pyramid is applied on the basis of the convolutional neural network to realize the multi-scale feature extraction and fusion of the image, which improves the matching similarity of the image blocks. The guide graph filter is used to quickly and effectively complete the cost aggregation. The disparity selection stage adapts the improvement dynamic programming algorithm to obtain the initial disparity map. The initial disparity map is refined so as to obtain the final disparity map. The algorithm is trained and tested on the image provided by Middlebury data set, and the result shows that the disparity map obtained by the algorithm has good effect. |
| Key words:
stereo matching
neural network
feature pyramid
multi-scale
guided image filter
dynamic programming
|
|
| 收稿日期: 2020-11-13
修回日期:
|
| DOI: 10.1051/jnwpu/20213940876 |
| 基金项目: 国家自然科学基金（51705051）、中国博士后科学基金（2019T120813，2018M643420）资助 |
| 通讯作者:
Email： |
| 作者简介: 陈星(1985-),重庆文理学院副教授,主要从事双目立体视觉及深度学习研究。
|
|
| 相关功能 |
|
|
|
|
| 作者相关文章 |
|
| 陈星 在本刊中的所有文章 |
| 张文海 在本刊中的所有文章 |
| 候宇 在本刊中的所有文章 |
| 杨林 在本刊中的所有文章 |
|
|
|
|
|
|
|
|
参考文献: |
|
|
[1] 张欢, 安利, 张强, 等. SGBM算法BM算法分析研究[J]. 测绘与空间地理信息, 2016, 39(10):214-216 ZHANG Huan, AN Li, ZHANG Qiang, et al. SGBM algorithm and BM algorithm analysis and research[J]. Geomatics and Spatial Information Technology, 2016, 39(10):214-216(in Chinese)
[2] JAC C K M, BYOUNG T. A dense stereo matching using two-pose dynamic programming with generalized ground control points[C]//IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2005
[3] BESSE F, ROTER C, FITZGIBBON A, et al. PMBP:patch match belief propagation for correspondence field estimation[J]. International Journal of Computer Vision, 2014, 110(1):2-13
[4] ŽBONTA J, LECUN Y. Stereo matching by training a convolutional neural network to compare image patches[J]. Journal of Machine Learning Research, 2016, 17(1):2287-2318
[5] SHAKED A, WOLF L. Improved stereo matching with constant highway networks and reflective confidence learning[C]//IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Washington, 2017
[6] 徐雪松,吴俊杰. 多尺度融合的立体匹配算法[J]. 模式识别与人工智能,2020,33(2):182-187 XU Xuesong, WU Junjie. Stereo matching algorithm based on multiscale fusion[J]. Patten Recognition and Artificial Intelligence, 2020, 33(2):182-187(in Chinese)
[7] 严邓涛. 基于卷积神经网络的立体匹配算法研究[D]. 南京:南京邮电大学,2019:39-50 YAN Dengtao. Research on stereo matching algorithm based on convolutional neural network[D]. Nanjing:Nanjing University of Posts and Telecommunications, 2019:39-50(in Chinese)
[8] LIN T, DOLLAR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]//IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2017
[9] SCHARSTEIN D, SELISKI R, ZABIH R. A Taxonomy and evaluation of dense two-frame stereo correspondence algorithms[C]//IEEE Stereo and Multi-Baseline Vision, 2002
[10] ZAGORUYKO S, KOMODAKIS N. Learning to compare image patches via convolutional neural networks[C]//IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2015
[11] SCHARSTEIN D, HIRSCHMVLLER H, KITAJIMA Y, et al. High-resolution stereo datasets with subpixel-accurate ground truth[C]//German Conference on Pattern Recognition, Berlin, 2014:31-42 |
|
|
|
相关文献: |
|
|
1．刘畅, 陈雯柏.一种基于MSDCNN-LSTM的设备RUL预测方法[J]. 西北工业大学学报, 2021,39(2): 407-413 |
|
|
|
|
|
|
