Three-dimensional structure from motion recovery of a moving object with noisy measurement

Abstract

In this paper, a Nonlinear Unknown Input Observer (NLUIO) based approach is proposed for three-dimensional (3-D) structure from motion identification. Unlike the previous studies that require prior knowledge of either the motion parameters or scene geometry, the proposed approach assumes that the object motion is imperfectly known and considered as an unknown input to the perspective dynamical system. The reconstruction of the 3-D structure of the moving objects can be achieved using just two-dimensional (2-D) images of a monocular vision system. The proposed scheme is illustrated with a numerical example in the presence of measurement noise for both static and dynamic scenes. Those results are used to clearly demonstrate the advantages of the proposed NLUIO.

Reconstruction of 3D Non-Rigid Moving Human Targets Based on Reliable Estimation of Contour Deformation Degree

“…In order to verify the optimization performance of the algorithm in this paper for the three-dimensional reconstruction of non-rigid moving human objects, we collected about 40 basketball body sequences of 2015 freshmen from a sports college, with 4 basketball positions sampled for each subset, and 10 positions for various positions. We used the algorithms in References [4,5,6,7,8] as the comparison algorithms of this algorithm, and used six algorithms to reconstruct the human body targets of basketball players in four movements, namely shooting, running, jumping, and passing. Combined with the convolutional neural network action recognition model, the reconstruction results of different algorithms are identified, and the action recognition errors of the reconstruction results of different algorithms are counted.…”

“…Method of reference [5] Method of reference [6] Method of this paper Method of reference [7] Method of reference [8] The experimental results from Figures 5-7 reveal that under the conditions of reconstructing different frames of images using different methods, the approach in Reference [4] frequently exhibits the phenomenon of missing information in certain body parts. This may be attributed to the method's inability to completely capture certain body part information during rapid human movement or complex postures, leading to occlusion or partial information loss and consequently resulting in poor reconstruction effectiveness.…”

“…Method of reference [5] Method of reference [6] Method of this paper Method of reference [7] Method of reference [8] The method in Reference [6] may have registration issues due to the assumption that the object's motion is not completely known, leading to inaccurate matching of feature point positions from different perspectives or the failure to accurately integrate information from certain regions into the overall reconstruction process, thus affecting the accuracy of the three-dimensional reconstruction.…”

“…Due to the involvement of multiple continuous images in registration, this method also adopts hierarchical registration, but the operation of this method is relatively complex, which affects the reliability and applicability of the algorithm. Mejri proposed a 3D motion recovery method for non-rigid objects based on noise measurement [6]. This method assumes that the motion of the object is not fully known and considers it as an unknown input to the perspective dynamic system, which leads to blurring in the reconstruction effect of this method.…”

Unknown input observer for Takagi-Sugeno implicit models with unmeasurable premise variables

“…This is especially due to its essential role in the fault diagnosis and fault tolerant control strategy development. Several research works using different approaches have been published in [17]- [20]. Many publications have been interested in the design of UIs for the case of T-S explicit models, we may refer to [21]- [26].…”