A Highly Efficient GPU Implementation for Variational Optic Flow Based on the Euler-Lagrange Framework

Gwosdek, Pascal; Zimmer, Henning; Grewenig, Sven; Bruhn, Andrés; Weickert, Joachim

doi:10.1007/978-3-642-35740-4_29

Cited by 27 publications

(21 citation statements)

References 21 publications

(60 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…By means of 2-D optic flow computations, it has already been demonstrated that FED is very well-suited for parallelisation on GPUs [13]. In this subsection, we illustrate that this also holds for Fast-Jacobi and in three dimensions.…”

Section: Higher Dimensional Problems and Gpu Implementationssupporting

confidence: 56%

See 1 more Smart Citation

Cyclic Schemes for PDE-Based Image Analysis

et al. 2015

Self Cite

View full text Add to dashboard Cite

The simplest scheme for parabolic, diffusion-like partial differential equations (PDEs) in image analysis is given by the explicit finite difference discretisation. If a fixed time step size is used, this scheme is known to be inefficient due to a severe stability restriction. In this paper we show that a slight modication can boost the efficiency of the explicit scheme by several orders of magnitude. All one has to do in this novel Fast Explicit Diffusion (FED) scheme is to replace the originally fixed time step size by cycles of varying step sizes. We derive these cycles from a factorisation of symmetric smoothing filters, and we show that a box filter gives a favourable compromise between high efficiency and good smoothing properties. However, such ideas are not restricted to parabolic problems only. They can also be adapted for solving elliptic PDEs that arise e.g. as Euler-Lagrange equations in variational image analysis. Also in this context, we propose a modification of the simplest iterative solver, namely the Jacobi method. This time, we introduce cycles of varying relaxation parameters that can be linked to our FED time step sizes. As a result, we obtain highly efficient methods for the elliptic case, so-called Fast-Jacobi (FJ) schemes. Our novel FED and FJ schemes are applicable to a large range of PDEbased image analysis problems and can beat classical methods such as additive operator splittings and (semi-)implicit schemes. Applications include isotropic nonlinear diffusion filters with widely varying diffusivities as well as anisotropic diffusion methods for image filtering, inpainting and regularisation in computer vision. Moreover, they are equally suited for higher dimensional problems as well as higher order PDEs. Implementations are extremely simple, since the underlying explicit scheme or Jacobi method can be used as a black box solver. Last but not least, these cyclic schemes are perfectly suited for modern parallel architectures such as GPUs. We also provide a public domain code package that allows users to experiment with our cyclic methods.

show abstract

Section: Higher Dimensional Problems and Gpu Implementationssupporting

confidence: 56%

“…for optic flow computation with a parallel GPU implementation [13], fast filtering methods on smartphones [14], medical image analysis [15,16], variational depth-from-defocus [17], and a cyclic projected gradient method for convex optimisation [18].…”

Section: Introductionmentioning

confidence: 99%

Cyclic Schemes for PDE-Based Image Analysis

et al. 2015

Self Cite

View full text Add to dashboard Cite

show abstract

“…This is due to the parallel treatment of the pixels within video frames, exploiting the large number of computing units in GPUs. We note also that we obtain reduced acceleration when processing low resolution 8 …”

Section: Performancementioning

confidence: 80%

“…Authors in [7] presented the CUDA implementation of the Horn-Shunck optical flow, that offered a real-time processing of 316×252 video resolution. Gwosdek et al [8] developed a GPU implementation of the Euler-Lagrange (EL) framework for solving variational optical flow methods using sequences with 640x480 pixels in near-real-time.…”

Section: Related Workmentioning

confidence: 99%

Multi-GPU based framework for real-time motion analysis and tracking in multi-user scenarios

Mahmoudi¹

2015

EAI Endorsed Transactions on Creative Technologies

View full text Add to dashboard Cite

Video processing algorithms present a necessary tool for various domains related to computer vision such as motion tracking, event detection and localization in multi-user scenarios (crowd videos, mobile camera, scenes with noise, etc.). However, the new video standards, especially those in high definitions require more computation since their treatment is applied on large video frames. As result, the current implementations, even running on modern hardware, cannot provide a real-time processing (25 frames per second, fps). Several solutions have been proposed to overcome this constraint, by exploiting graphic processing units (GPUs). Although they exploit GPU platforms, they are not able to provide a real-time processing of high definition video sequences. In this work, we propose a new framework that enables an efficient exploitation of single and multiple GPUs, in order to achieve real-time processing of Full HD or even 4K video standards. Moreover, the framework includes several GPU based primitive functions related to motion analysis and tracking methods, such as silhouette extraction, contours extraction, corners detection and tracking using optical flow estimation. Based on this framework, we developed several real-time and GPU based video processing applications such as motion detection using moving camera, event detection and event localization

show abstract

“…Another approach is to consider the solution of the Euler-Lagrange equation as the steady state of the corresponding diffusion process (34), and use the Fast Explicit Diffusion (FED) principle [105] to accelerate convergence by adapting the time steps. The implementation of [107] exploits the natural parallelization of explicit schemes to achieve a quasi real-time version of the variational method of [286] on GPU, based on FED.…”

Section: Continuous Methodsmentioning

confidence: 99%

Optical flow modeling and computation: A survey

Fortun

Bouthémy

Kervrann

2015

Computer Vision and Image Understanding

353

182

View full text Add to dashboard Cite

a b s t r a c tOptical flow estimation is one of the oldest and still most active research domains in computer vision. In 35 years, many methodological concepts have been introduced and have progressively improved performances, while opening the way to new challenges. In the last decade, the growing interest in evaluation benchmarks has stimulated a great amount of work. In this paper, we propose a survey of optical flow estimation classifying the main principles elaborated during this evolution, with a particular concern given to recent developments. It is conceived as a tutorial organizing in a comprehensive framework current approaches and practices. We give insights on the motivations, interests and limitations of modeling and optimization techniques, and we highlight similarities between methods to allow for a clear understanding of their behavior.

show abstract

A Highly Efficient GPU Implementation for Variational Optic Flow Based on the Euler-Lagrange Framework

Cited by 27 publications

References 21 publications

Cyclic Schemes for PDE-Based Image Analysis

Cyclic Schemes for PDE-Based Image Analysis

Multi-GPU based framework for real-time motion analysis and tracking in multi-user scenarios

Optical flow modeling and computation: A survey

Contact Info

Product

Resources

About