基于相位相关和纹理分类的SIFT
图像拼接算法

量子电子学报 ›› 2020, Vol. 37 ›› Issue (6): 650-658.

基于相位相关和纹理分类的SIFT 图像拼接算法

王昱皓, 唐泽恬, 钟岷哲, 王阳, 曾瑞敏, 朱登玮, 杨晨

贵州大学大数据与信息工程学院半导体功率器件可靠性教育部工程研究中心, 贵州贵阳550025

收稿日期:2020-01-17 修回日期:2020-05-08 出版日期:2020-11-28 发布日期:2020-11-28
通讯作者: E-mail: eliot.c.yang@163.com
作者简介:王昱皓( 1992 - ), 河南开封人, 研究生, 主要从事图像处理和机器视觉方面的研究。E-mail: wyh920726@163.com
基金资助:
Supported by National Natural Science Foundation of China (国家自然科学基金, 61604046), Guizhou Science and Technology Plan Project (贵州省科技计划项目, 黔科合平台人才[2017]5788号, [2018]5781号), Semiconductor Power Device Reliability Ministry of Education Engineering Research Center Open Fund (半导体功率器件可靠性教育部工程研究中心开放基金, 黔科合平台人才20176103号)

SIFT image stitching algorithm based on phase correlation and texture classification

WANG Yuhao, TANG Zetian, ZHONG Minzhe, WANG Yang, ZENG Ruimin, ZHU Dengwei, YANG Chen

Power Semiconductor Device Reliability Engineering Center of the Ministry of Education, College of Big Data and Information Engineering, Guizhou University, Guiyang 550025, China

Received:2020-01-17 Revised:2020-05-08 Published:2020-11-28 Online:2020-11-28

摘要/Abstract

摘要： 针对传统尺度不变特征变换(SIFT) 算法在图像拼接过程中计算量较大的问题, 提出了一种基于相位相关和纹理分类的快速SIFT 拼接算法。首先, 使用相位相关法粗略地获取待拼接图像的重叠区域; 其次, 对图像进行纹理分类, 并选取其中纹理复杂度较高区域进行SIFT 检测, 为了提升纹理分类阶段的速度, 提出了跳跃归类的方法; 最后, 对于不同复杂度的纹理区域, 仅在相同纹理区域内进行特征点的匹配。实验结果表明, 所提出改进算法在维持较好拼接质量的基础上, 与传统SIFT 算法以及已有的两种改进型SIFT 算法相比, 平均拼接速度分别提升了68.46%、20.45%、41.83%。因此该算法在对拼接效率有较高要求的领域具有潜在的应用价值。

关键词: 图像处理, 图像拼接, 尺度不变特征变化, 相位相关, 纹理分类, 跳跃归类

Abstract: Aiming at the problem that the traditional scale-invariant feature transform (SIFT) algorithm has a large amount of calculation in the process of image stitching, a fast SIFT stitching algorithm based on phase correlation and texture classification is proposed. Firstly, by using the phase correlation method, the overlapping regions of the input images to be stitched are roughly obtained. Secondly, images are classified with texture and the area with higher texture complexity is selected for SIFT detection. In order to improve the speed of texture classification, an interval classification approach is proposed. Finally, the feature points are matched only in the regions with the same texture complexity for different complexity texture regions. The experimental results show that compared with the traditional SIFT algorithm and the two existing improved SIFT algorithms, the improved algorithm in this work not only maintains good stitching quality, but also improves the average stitching speed by 68.46%, 20.45%, 41.83%, respectively. Therefore, the proposed algorithm has the potential application value in the field of high stitching efficiency requirements.

Key words: image processing, image stitching, scale-invariant feature transform, phase correlation; texture classification, interval classification

中图分类号:

TP391.41

王昱皓, 唐泽恬, 钟岷哲, 王阳, 曾瑞敏, 朱登玮, 杨晨. 基于相位相关和纹理分类的SIFT 图像拼接算法[J]. 量子电子学报, 2020, 37(6): 650-658.

WANG Yuhao, TANG Zetian, ZHONG Minzhe, WANG Yang, ZENG Ruimin, ZHU Dengwei, YANG Chen. SIFT image stitching algorithm based on phase correlation and texture classification[J]. Chinese Journal of Quantum Electronics, 2020, 37(6): 650-658.

参考文献 18

[1]	Ghosh D, Kaabouch N. A survey on image mosaicing techniques [J]. Journal of Visual Communication & Image Representation,
20	16, 34(C): 1-11.
[2]	Xiao Ming, Bao Yongliang, Yan Zhongxin. Point feature-based image registration: A survey [J]. Acta Armamentarii (兵工学
报)	, 2015, 36(S2): 326-340 (in Chinese).
[3]	Tang Zetian, Ding Zhao, Zeng Ruimin, et al. Quantum image stitching of SIFT based on dynamic threshold and global
	information [J]. Chinese Journal of Quantum Electronics (量子电子学报), 2020, 37(1): 22-28 (in Chinese).
[4]	Lu Peng, Lu Qi, Zou Guoliang, et al. Research on time series image mosaic method based on improved SIFT [J]. Computer
	Engineering and Applications (计算机工程与应用), 2020, 56(1): 196-202 (in Chinese).
[5]	Ma W P, Wen Z L, Wu Y, et al. Remote sensing image registration with modified SIFT and enhanced feature matching [J].
	IEEE Geoscience and Remote Sensing Letters, 2016, 14(1): 3-7.
[6]	Lowe D G. Object reconition from local scale-invarriant features [C]. The Seventh IEEE International Conference on Computer
	Vision Kerkyra, 1999: 1150-1157.
[7]	Lowe D G. Distinctive image features from scale-invariant keypoints [J]. International Journal of Computer Vision, 2004,
60	(2): 91-110.
[8]	Sun Xueqiang, Huang Min, Zhang Guifeng, et al. Multispectral image matching algorithm based on improved SIFT [J].
	Computer Science (计算机科学), 2019, 46(4): 280-284 (in Chinese).
[9]	Chang H H,Wu G L, Chiang M H. Remote sensing image registration based on modified SIFT and feature slope grouping [J].
	IEEE Geoscience and Remote Sensing Letters, 2019, 16(9): 1363-1367.
[10]	Li Yufeng, Li Guangze, Gu Shaohu, et al. Image mosaic algorithm based on area blocking and SIFT [J]. Optics and Precision
	Engineering (光学精密工程), 2016, 24(5): 1197-1205 (in Chinese).
[11]	Ke Y, Sukthankar R. PCA-SIFT: A more distinctive representation for local image descriptors [C]. IEEE Computer Society
	Conference on Computer Vision & Pattern Recognition, 2004: 506-513.
[12]	He Bin, Tao Dan, Peng Bo. High real-time F-SIFT image mosaic algorithm [J]. Infrared and Laser Engineering (红外与激光
工程)	, 2013, 42(S2): 440-444 (in Chinese).
[13]	Xu Jiajia. Fast image registration method based on Harris and SIFT algorithm [J]. Chinese Optics (中国光学), 2015, 8(4):
57	4-581 (in Chinese).
[14]	Yang Sa, Zheng Zhishuo. Medical image registration algorithm based on sparse random projection and SIFT transform [J].
	Chinese Journal of Quantum Electronics (量子电子学报), 2015, 32(3): 283-289 (in Chinese).
[15]	Dong Jintao, Chen Zhongshui, Xu Kai, et al. A real-time improvement of image matching based on SIFT [J]. Electronics
	Optics & Control (电光与控制), 2020, 27(3): 80-83 (in Chinese).
[16]	Han Yu, Zong Qun, Xing Na. Fast matching of UAV aerial image based on SIFT [J]. Acta Scientiarum Naturalium Universitatis
	Nankaiensis (南开大学学报(自然科学版)), 2019, 52(1): 5-9 (in Chinese).
[17]	Yang Lei, Cao Jianzhong, Tang Linao, et al. Optimized design of automatic panoramic images mosaic [J]. Infrared and Laser
	Engineering (红外与激光工程), 2014, 43(3): 985-990 (in Chinese).
[18]	Liu Wei, Tang Cunchen. Medical image fusion algorithm based on salient features coupling local extreme decomposition [J].
	Chinese Journal of Quantum Electronics (量子电子学报), 2018, 35(2): 156-164 (in Chinese).