Stereo vision-based variational optical flow estimation

Optical flow computation is one of the most important tasks in computer vision. The article deals with a modification of the variational method of the optical flow computation, according to its application in stereo vision. Such approaches are traditionally based on a brightness constancy assumption and a gradient constancy assumption during pixels motion. Smoothness assumption also restricts motion discontinuities, i.e. the smoothness of the vector field of pixel velocity is assumed. It is proposed to extend the functional of the optical flow computation in a similar way by adding a priori known stereo cameras extrinsic parameters and minimize such jointed model of optical flow computation. The article presents a partial differential equations framework in image processing and numerical scheme for its implementation. Performed experimental evaluation demonstrates that the proposed method gives smaller errors than traditional methods of optical flow computation.

Occlusion-aware temporal frame interpolation in a highly scalable video coding setting

We recently proposed a bidirectional hierarchical anchoring (BIHA) of motion fields for highly scalable video coding. The BIHA scheme employs piecewise-smooth motion fields, and uses breakpoints to signal motion discontinuities. In this paper, we show how the fundamental building block of the BIHA scheme can be used to perform bidirectional, occlusion-aware temporal frame interpolation (BOA-TFI). From a “parent” motion field between two reference frames, we use information about motion discontinuities to compose motion fields from both reference frames to the target frame; these then get inverted so that they can be used to predict the target frame. During the motion inversion process, we compute a reliable occlusion mask, which is used to guide the bidirectional motion-compensated prediction of the target frame. The scheme can be used in any state-of-the-art codec, but is most beneficial if used in conjunction with a highly scalable video coder which employs piecewise-smooth motion fields with motion discontinuities. We evaluate the proposed BOA-TFI scheme on a large variety of natural and challenging computer-generated sequences, and our results compare favorably to state-of-the-art TFI methods.

Tracking Contrast Agents from Ultrasound Image Using Texture Preserving Optical Flow Estimation

In this paper, we proposed a novel texture preserving optical flow technique to estimate the motion patterns of contrast agent on the ultrasound image. The proposed method estimated the motion fields based on three major steps. Firstly, the proposed method recomposed the original image based on the weighted structure-texture decomposition. Secondly, we applied a slightly non-convex approximation approach by utilizing the spline interpolation based coarse-to-fine warping scheme to handle the motion discontinuities in ultrasound image. Lastly, after each warping step, we employed the bilateral filter into the numerical framework to minimize the tracking errors in motion estimates. To evaluate the tracking performance of our method, we estimated the motion fields of microbubbles for the tissue mimicking phantom, and compared its results to those of the existing methods. As a result, it was found that the proposed technique provides the most reliable motion patterns of microbubbles, and reduces computational loads, simultaneously. We also confirmed the potential usefulness of our estimation scheme for the microbubble based diagnostic analysis.

Real-time Optical-flow Computation for Motion Estimation under Varying Illumination Conditions

<p>The optical flow approach has emerged as a major technique for estimating object motion in image sequences. However, the obtained results by most optical flow techniques are poor because they are strongly affected by large illumination changes and by motion discontinuities. On the other hand, there have been two thrusts in the development of optical flow algorithms. One has emphasized higher accuracy; the other faster implementation. These two thrusts have been independently pursed, without addressing the accuracy vs. efficiency trade-offs. The optical flow computation requires high computing resources and is highly affected by changes in the illumination conditions in most of the existing techniques. In this paper, a new strategy for image sequence processing is proposed. The data reduction achieved with this strategy allows a faster optical flow computation. In addition, the proposed architecture is a hardware custom implementation in EP1S60F1020 FPGA showing the achieved performance.</p>

Directionally Weakened Diffusion for Optical Flow

Optical flow estimation from image sequences is of paramount importance to computer vision applications. Many optical flow algorithms have been proposed for optical flow computation, which minimize a certain energy function involving a data term and a smoothness term. In this paper, we propose a new method to compute optical flow by decomposing the Laplacian operator along the tangential and gradient direction of the image. Experimental results show that the new method could gain more accurate estimation of optical flow around motion discontinuities compared with the classical methods.

A Robust Multiscale Algorithm for Motion Measurement With High Accuracy

The computer micro-vision technique plays a vital role in many existing methods of motion measurement in precision engineering, which its measurement accuracy and speed perform very well. Gradient-based techniques represent a very popular class of approaches to measuring motions. A robust multiscale algorithm of hierarchical estimation for gradient-based motion estimation is proposed in this paper using a combination of robust statistical method and multiscale technique. In such a multiscale approach of hierarchical estimation, motion at each level of the pyramid is estimated using different gradient filters. The iterative multiscale estimation begins by using 5-tap Central filter, and it is switched to 9-tap Timoner filter after a few iterations. In addition, robust M-estimators are applied at each level of the pyramid in order to overcome the problem of the outliers caused by illumination variations and motion discontinuities in motion estimates. Experimental simulations show that the new algorithm not only provides an improvement in estimator accuracy, but also achieves computational speedups for motion measurement of MEMS image.