Shape point set matching based on oriented shape context in turbulence-cluttered scene
- PDF / 2,480,514 Bytes
- 18 Pages / 439.37 x 666.142 pts Page_size
- 41 Downloads / 151 Views
Shape point set matching based on oriented shape context in turbulence-cluttered scene Xinggui Xu 1,2
3
3
& Ping Yang & Hao Xian & Yong Liu
2
Received: 30 October 2019 / Revised: 4 June 2020 / Accepted: 12 June 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
The main challenges of shape point set matching in a long-distance imaging scene stem from the optical turbulence effects, which lead to shape object deformation, rotation, shifted object positions, and cluttered outliers. To address this problem, we propose an effective energy cost model with figural continuity constraint. We first construct an Oriented Shape Context (OSC) descriptor using attributes of shape edges’ length and direction, which represent rotation invariance, by adding the oriented model (prototype) edges point set. Then, inspired by the figural continuity prior between the model and target point set, we transform the continuity constraint into a matching energy cost model. Lastly, we develop a simple 2-tree graph to minimize the matching cost function using the Dynamic Program (DP) optimization algorithm. The extensive experiments on both synthetic and real data validate that the proposed method can effectively detect the desired shapes in the complex and highly turbulence-cluttered scenes. Keywords Imaging through turbulent media . Shape invariant descriptor . Cluttered scene . Shape context . Contour shape recognition
1 Introduction Image contour shape is a vital visual characteristic that can be used to recognize objects with a lack of texture and color. The usual procedure for object recognition is to match two or more
* Xinggui Xu [email protected]
1
Department of Physics, Center for Optoelectronics Engineering Research, Yunnan University, Kunming 650091, China
2
School of Optoelectronic Science and Engineering, University of Electronic Science and Technology of China, Chengdu 610054, China
3
Key Laboratory on Adaptive Optics and Institute of Optics and Electronics, Chinese Academy of Sciences, Chengdu 610209, China
Multimedia Tools and Applications
point sets that samples from contour shape or image. However, there are still challenges of shape point set matching in a long-distance imaging system (LDIS) [23] that stem from the optical turbulence effects, which lead to shape object deformation, shifted object positions, and outliers. To effectively address this matching problem, the development of a shape correspondence point’s model (prototype) and the transformation are the main motivation, as presented in [10, 11, 21]. Especially, the better the discrimination ability of a shape descriptor is, the better the point set correspondence transformation matching can be achieved, and vice versa. In fact, some advanced features extraction methods, such as Scale-invariant Feature Transform (SIFT) [13], Shape Context (SC) [4], Inner Distance Shape Context (IDSC) [12], cannot remove all the deformation and noise, which largely degrades recognition performance. In practice, a long-distance turbu
Data Loading...
