基于光照模糊相似融合不变矩的航天器目标识别
收稿日期: 2013-07-12
修回日期: 2013-09-28
网络出版日期: 2013-10-10
基金资助
国家自然科学基金（60974105，61104188）；航空科学基金（20100152003）
CIBA Moment Invariants and Their Use in Spacecraft Recognition Algorithm
Received date: 2013-07-12
Revised date: 2013-09-28
Online published: 2013-10-10
Supported by
National Natural Science Foundation of China (60974105,61104188); Aeronautical Science Foundation of China (20100152003)
针对航天器目标识别中现有的光照变化和图像模糊条件下几何仿射不变矩识别不稳健问题，根据灰度光照模型，结合图像模糊和不变矩的基本理论，提出了一种光照模糊不变量，与几何仿射不变矩进行融合，构建出一种新的光照模糊相似融合（CIBA）不变矩，并从理论上证明了光照、模糊和相似变换（平移、缩放和旋转）不变性。对3种不同的航天器目标的识别实验表明，光照模糊相似融合不变矩在最小距离分类器中的识别准确率达到了86.37%，比几何仿射不变矩提高了94.00%，比具有光照不变性的几何仿射不变矩提高了24.54%，有效地解决了对不同位姿、不同光照模糊条件下的航天器目标识别问题，提高了基于矩的目标识别鲁棒性。
徐贵力 , 徐静 , 王彪 , 田裕鹏 , 郭瑞鹏 , 吕温 . 基于光照模糊相似融合不变矩的航天器目标识别[J]. 航空学报, 2014 , 35(3) : 857 -867 . DOI: 10.7527/S1000-6893.2013.0403
In order to solve the non-robustness for affine moment invariants with changing illumination conditions and unknown blurring, an illumination and blur invariant is proposed which is based on the gray light model, image blurring theory and basic moment invariants theory. Combining the geometric affine moment invariants with the proposed illumination and blur invariants, the combined illumination, blur and affine (CIBA) moment invariants are built and their invariant features of illumination, blur and similarity transform (i.e. rotation, scaling and translation) are theoretically proved. Taking multiple images of three spacecraft models as examples, the average recognition accuracy by using CIBA moment invariants in a linear minor distance classifier can reach 86.37%, which is a 94.00% improvement compared with the affine moment invariants, and a 24.54% improvement as compared with the illumination moment invariants. CIBA moment invariants can effectively solve the problem in the identification of spacecraft under different conditions of illumination, unknown blurring and pose. They enhance the target recognition robustness which is based on moment invariants.
[1] Wang X X, Yang Y S, Jing Z L. Space target recognition based on improved kernel FCM[J]. Chinese Space Science and Technology, 2012, 32(2): 35-42. (in Chinese) 王晓雪, 杨永胜, 敬忠良. 基于改进核聚类算法的空间目标识别方法[J]. 中国空间科学技术, 2012, 32(2): 35-42.
[2] Li L R, Shen C L, Wang C Q. Recognition method of aerial targets based on BP neural network[J]. Fire Control & Command Control, 2012, 37(12): 132-139. (in Chinese) 李丽荣, 沈春林, 王从庆. 基于BP神经网络的空中目标识别方法[J]. 火力与指挥控制, 2012, 37(12): 132-139.
[3] Flusser J, Suk T. Degraded image analysis: an invariant approach[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1998, 20(6): 590-603.
[4] Suk T, Flusser J. Combined blur and affine moment invariants and their use in pattern recognition[J]. Pattern Recognition, 2003, 36(12): 2895-2907.
[5] Ji H J, Zhu H Q. Degraded image analysis using zernike moment invariants[C]//IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2009: 1941-1944.
[6] Liu J. Construction of feature invariants and its application in object recognition[D]. Wuhan: Huazhong University of Science and Technology, 2004. (in Chinese) 刘进. 不变量特征的构造及在目标识别中的应用[D]. 武汉: 华中科技大学, 2004.
[7] Dai X B, Zhang H, Shu H Z, et al. Image recognition by combined invariants of Legendre moment[C]//Proceedings of the 2010 IEEE International Conference on Information and Automation, 2010: 1793-1798.
[8] Paschalakis S, Lee P. Combined geometric transformation and illumination invariant object recognition in RGB color images[C]//Proceedings of 15th International Conference on Pattern Recognition, 2000: 584-587.
[9] Mindru F, Tuytelaars T, van Gool L, et al. Moment invariants for recognition under changing viewpoint and illumination[J]. Computer Vision and Image Understanding, 2004, 94(1-3): 3-27.
[10] Suk T, Flusser J. Affine moment invariants of color images[M]//Computer Analysis of Images and Patterns. New York: Springer, 2009: 334-341.
[11] Lu Y. Stability of moment invariants and their applications in 3-D object recognition[D]. Wuhan: Huazhong University of Science and Technology, 2007. (in Chinese) 路鹰. 不变矩稳定性及在三维目标识别中的应用[D].武汉: 华中科技大学, 2007.
[12] Liu J, Zhang T X. Fast algorithm for generation of moment invariants[J]. Pattern Recognition, 2004, 37(8): 1745-1756.
[13] Wang Y P, Huang X S, Li X L, et al. Moment invariants of gray distorted based on Hu Moments[J]. Journal of Dalian Maritime University, 2008, 34(4): 23-27. (in Chinese) 王亦平, 黄新生, 李晓雷, 等. 一种新的灰度变化不变矩[J]. 大连海事大学学报, 2008, 34(4): 23-27.
[14] Xu G L, Zhong Z W, Wang B, et al. Research on the construction of moment invariants of the geometry and illumination invariance[J]. Acta Aeronautica et Astronautica Sinica, 2013, 34(7): 1698-1705. (in Chinese) 徐贵力, 钟志伟, 王彪, 等. 具有几何和光照不变性的不变矩构建研究[J]. 航空学报, 2013, 34(7): 1698-1705.
[15] Gong M, Hu P, Cao W G, et al. A kind of shape-color moment invariants[C]//International Conference on Computer-Aided Design and Computer Graphics, 2011: 425-432.
[16] Hu M K. Visual pattern recognition by moment invariants[J]. IRE Transactions on Information Theorem, 1962, 8(2): 179-187.
[17] Suk T, Flusser J. Affine moment invariants generated by graph method[J]. Pattern Recognition, 2011, 44(9): 2047-2056.
[18] Gevers T, Smeulders A W M. Color-based object recognition[J]. Pattern Recognition, 1999, 32(3): 453-464.
/
| 〈 |
|
〉 |
CJA