An Adaptive Confidence Measure for Optical Flows Based on Linear Subspace Projections

Kondermann, Claudia; Kondermann, Daniel; Jähne, Bernd; Garbe, Christoph S.

doi:10.1007/978-3-540-74936-3_14

Cited by 37 publications

(22 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There have been many attempts to derive useful confidence measures for OF methods, the main application being to identify unreliable flow vectors for error statistics reporting [4] and weighting or pruning for subsequent processing steps [21,22].…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Bootstrap optical flow confidence and uncertainty measure

Kybic

Nieuwenhuis

2011

Computer Vision and Image Understanding

View full text Add to dashboard Cite

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues.Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. a b s t r a c tWe address the problem of estimating the uncertainty of optical flow algorithm results. Our method estimates the error magnitude at all points in the image. It can be used as a confidence measure. It is based on bootstrap resampling, which is a computational statistical inference technique based on repeating the optical flow calculation several times for different randomly chosen subsets of pixel contributions. As few as ten repetitions are enough to obtain useful estimates of geometrical and angular errors. For demonstration, we use the combined local-global optical flow method (CLG) which generalizes both LucasKanade and Horn-Schunck type methods. However, the bootstrap method is very general and can be applied to almost any optical flow algorithm that can be formulated as a pixel-based minimization problem. We show experimentally on synthetic as well as real video sequences with known ground truth that the bootstrap method performs better than all other confidence measures tested.

show abstract

Section: Related Workmentioning

confidence: 99%

“…The linear subspace of the 'correct' flow can be learnt by PCA [21]. Alternatively, the dependence of the flow vectors in a patch on the central vector can be modeled as a multidimensional Gaussian [22].…”

Section: Related Workmentioning

confidence: 99%

Bootstrap optical flow confidence and uncertainty measure

Kybic

Nieuwenhuis

2011

Computer Vision and Image Understanding

View full text Add to dashboard Cite

show abstract

“…These incorrect flow patterns can be detected and removed from the flow field e.g. by means of confidence measures [1][2][3]. But since many applications demand a dense flow field, it would be beneficial to reconstruct a reliable dense vector field based on information from the surrounding flow field.…”

Section: Introductionmentioning

confidence: 99%

Reconstructing Optical Flow Fields by Motion Inpainting

Berkels

Kondermann

Garbe

et al. 2009

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Abstract. An edge-sensitive variational approach for the restoration of optical flow fields is presented. Real world optical flow fields are frequently corrupted by noise, reflection artifacts or missing local information. Still, applications may require dense motion fields. In this paper, we pick up image inpainting methodology to restore motion fields, which have been extracted from image sequences based on a statistical hypothesis test on neighboring flow vectors. A motion field inpainting model is presented, which takes into account additional information from the image sequence to improve the reconstruction result. The underlying functional directly combines motion and image information and allows to control the impact of image edges on the motion field reconstruction. In fact, in case of jumps of the motion field, where the jump set coincides with an edge set of the underlying image intensity, an anisotropic TV-type functional acts as a prior in the inpainting model. We compare the resulting image guided motion inpainting algorithm to diffusion and standard TV inpainting methods.

show abstract

“…Particularly in the presence of strong noise, structure tensor methods requires several successive images for recovering highly accurate gradients. For the present application, a local approach would thus lead to noisy motion fields or very sparse ones, if inaccurate flows are excluded by using confidence measures (confer Kondermann et al 2007).…”

Section: Fluorescence Motion Analysis (Fma)mentioning

confidence: 99%

Double-pulse planar-LIF investigations using fluorescence motion analysis for mixture formation investigation

et al. 2008

View full text Add to dashboard Cite

A concept for dynamic mixture formation investigations of fuel/air mixtures is presented which can equally be applied to several other laser induced fluorescence (LIF) applications. Double-pulse LIF imaging was used to gain insight into dynamic mixture formation processes. The setup consists of a modified standard PIV setup. The ''fuel/air ratio measurement by laser induced fluorescence (FARLIF)'' approach is used for a quantification of the LIF images in order to obtain pairs of 2D fuel/air ratio maps. Two different evaluation concepts for LIF double pulse images are discussed. The first is based on the calculation of the temporal derivative field of the fuel/air ratio distribution. The result gives insight into the dynamic mixing process, showing where and how the mixture is changing locally. The second concept uses optical flow methods in order to estimate the motion of fluorescence (i.e., mixture) structures to gain insight into the dynamics, showing the distortion and the motion of the inhomogeneous mixture field. For this ''fluorescence motion analysis'' (FMA) two different evaluation approaches-the ''variational gradient based approach'' and the ''variational cross correlation based approach''-are presented. For the validation of both, synthetic LIF image pairs with predefined motion fields were generated. Both methods were applied and the results compared with the known original motion field. This validation shows that FMA yields reliable results even for image pairs with low signal/noise ratio. Here, the ''variational gradient based approach'' turned out to be the better choice so far. Finally, the experimental combination of doublepulse FARLIF imaging with FMA and simultaneous PIV measurement is demonstrated. The comparison of the FMA motion field and the flow velocity field captured by PIV shows that both results basically reflect complementary information of the flow field. It is shown that the motion field of the fluorescence structures does not (necessarily) need to represent the actual flow velocity and that the flow velocity field alone can not illustrate the structure motion in any case. Therefore, the simultaneous measurement of both gives the deepest insight into the dynamic mixture formation process. The examined concepts and evaluation approaches of this paper can easily be adapted to various other planar LIF methods (with the LIF signal representing, e.g., species concentration, temperature, density etc.) broadening the insight for a wide range of different dynamic processes.

show abstract

An Adaptive Confidence Measure for Optical Flows Based on Linear Subspace Projections

Cited by 37 publications

References 13 publications

Bootstrap optical flow confidence and uncertainty measure

Bootstrap optical flow confidence and uncertainty measure

Reconstructing Optical Flow Fields by Motion Inpainting

Double-pulse planar-LIF investigations using fluorescence motion analysis for mixture formation investigation

Contact Info

Product

Resources

About