Multi-nonlinear multi-view locality-preserving projection with similarity learning for random cross-view gait recognitio
- PDF / 3,728,345 Bytes
- 18 Pages / 595.276 x 790.866 pts Page_size
- 77 Downloads / 155 Views
REGULAR PAPER
Multi‑nonlinear multi‑view locality‑preserving projection with similarity learning for random cross‑view gait recognition Xiaoyun Chen1 · Yeyuan Kang1 · Zhiping Chen1 Received: 26 August 2019 / Accepted: 15 August 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract View variation is one of the greatest challenges in the field of gait recognition. Subspace learning approaches are designed to solve this issue by projecting cross-view features into a common subspace before recognition. However, similarity measures are data-dependent, which results in low accuracy when cross-view gait samples are randomly arranged. Inspired by the recent developments of data-driven similarity learning and multi-nonlinear projection, we propose a new unsupervised projection approach, called multi-nonlinear multi-view locality-preserving projections with similarity learning ( M2LPP-SL). The similarity information among cross-view samples can be learned adaptively in our M2LPP-SL. Besides, the complex nonlinear structure of original data can be well preserved through multiple explicit nonlinear projection functions. Nevertheless, its performance is largely affected by the choice of nonlinear projection functions. Considering the excellent ability of kernel trick for capturing nonlinear structure information, we further extend M 2LPP-SL into kernel space, and propose its multiple kernel version MKMLPP-SL. As a result, our approaches can capture linear and nonlinear structure more precisely, and also learn similarity information hidden in the multi-view gait dataset. The proposed models can be solved efficiently by alternating direction optimization method. Extensive experimental results over various view combinations on the multi-view gait database CASIA-B have demonstrated the superiority of the proposed algorithms. Keywords Subspace learning · Multiple kernels learning · Similarity learning · Multiple nonlinear projection · Locally preserving projection
1 Introduction Biometric emerged as an efficient status authentication technique that includes fingerprints, face, iris, retina, palm, DNA, gait, sound, etc. [1]. Multi-view learning technique is often used for face and gait recognition. For face recognition tasks, an image could be represented by different descriptors, such as SIFT, HOG, LBP, etc. [2]. Each descriptor of the same image [3] or each pose of the same person [4] can be regarded as a view. The multi-view recognition methods that use multiple descriptor features result in better accuracy because they indicate different aspects of the same image and offer complementary information. Communicated by B. Prabhakaran. * Xiaoyun Chen [email protected] 1
College of Mathematics and Computer Science, Fuzhou University, Fuzhou 350116, China
With the development of video surveillance system, gait recognition has received considerable attention over the last decade. Different from face image, multi-view gait samples can be captured at a distance from different positions of the camera.
Data Loading...
