High-Speed and Robust Scene Matching Algorithm Based on ORB for SAR/INS Integrated Navigation System

Abstract:

It is important that scene matching algorithm should satisfy the requirements of real-time, robustness and high-precision for inertial integrated navigation system. And considering the serious distortion and speckle noises of SAR images, we proposed a new scene matching algorithm for the SAR/INS integrated navigation system with high-speed and robustness based on Oriented FAST and Rotated BRIEF (ORB). We started by detecting scale-space FAST-based features in combination with an efficiently computed orientation in the image. Then, we calculated feature point's Rotation-Aware BRIEF descriptor which performs well with rotation and match features by computing Hamming distance between descriptors. Finally, we adopted GroupSAC which are proposed recently to remove the false matching points and the least square algorithm for getting the distortion transformation parameters that are the aircraft position errors and rotation transform parameters between real image and reference image. Experimental results on real SAR images indicate that our algorithm is invariant to various image transformations due to rotation and scale, and also robust to speckle noise and extremely efficient to compute, better than SIFT in many situations. Therefore, our algorithm can meet the high performance needs for matching navigation in the SAR/INS integrated navigation system.

References

[1] Liu Jianye, Leng Xuefei, Xiong Zhi, Li Mingxin. Real-time Multi-level Scene Matching Algorithm for Inertial Integrated Navigation System [J]. Acta Aeronautica Et Astronautica Sinica, 2007, 6 (28): 1401 – 1407(In Chinese).

[2] XIONG Zhi, LIU Jian-ye, ZENG Qing-hua, et al. The Study of Image Matching Algorithm for Scene Matching Aided Navigation System[J]. Journal of Image and Graphics, 2004, 9(1): 29-34 (In Chinese).

[3] CHEN Fang, XIONG Zhi, XU Yun-xi, et al. Research on the Fast Scene Matching Algorithm in the Inertial Integrated Navigation System[J]. Journal of Astronautics, 2009, 6(30): 2308-: 2316 (In Chinese).

[4] David G. Lowe. Object recognition from local scale-invariant features [J]. International Conference on Computer Vision, Corfu, Greece(September 1999): 1150-1157.

DOI: 10.1109/iccv.1999.790410

[5] David G. Lowe. Distinctive Image Features from Scale-Invariant Keypoints[J]. International Journal of Computer Vision, 2004, 2(60): 91-110.

DOI: 10.1023/b:visi.0000029664.99615.94

[6] Ethan Rublee, Vincent Rabaud, Kurt Konolige, et al. ORB: an Efficient Alternative to SIFT or SURF[C]. In Proceedings of the Twelfth International Conference on Computer Vision(ICCV), (2011).

DOI: 10.1109/iccv.2011.6126544

[7] M. Calonder, V. Lepetit, C. Strecha, et al. BRIEF: Binary Robust Independent Elementary Features[C]. European Conference on Computer Vision, Heraklion, Greece, (2010).

DOI: 10.1007/978-3-642-15561-1_56

[8] Ni K, Jin H, and Dellaert F. GroupSAC: Efficient Consensus in the Presence of Groupings[C]. In Proceedings of the Twelfth International Conference on Computer Vision(ICCV), (2009).

DOI: 10.1109/iccv.2009.5459241

[9] Hua Xie, L.E. Pierce, F.T. Ulaby. SAR Speckle Reduction Using Wavelet Denoising and Markov Random Field Modeling [J]. IEEE Trans. Geosciences And Remote Sensing, 2002, 10(4): 2196-2211.

DOI: 10.1109/tgrs.2002.802473