Fabian E. Ernst

Removal of scattered guided waves from seismic data

Near-surface scattered waves form a major source of coherent noise in seismic land data. Most current methods for removing these waves do not attenuate them adequately if they come from other than the inline direction. We present a wave-theory-based method for removing (scattered) guided waves by a prediction-and-removal algorithm. We assume that the near surface consists of a laterally varying medium, in which heterogeneities are embedded that act as scatterers. We first estimate the dispersive and laterally varying phase slowness field by applying a phase-based tomography algorithm on the direct groundroll wave. Subsequently, the near-surface heterogeneities are imaged using a least-squares criterium. Finally, the scattered guided waves are modeled and subtracted adaptively from the seismic data. We have applied this method to seismic land data and found that near-surface scattering effects are attenuated.

Comparison of shot boundary detectors

A video cut detector (CD), a member of the shot boundary detector (SBD) group, is an essential element for spatio-temporal audiovisual (AV) segmentation and various video-processing technologies. Platform, processing and performance constraints forced the development of various dedicated CDs. Future platforms allow the usage of advanced CD algorithms with higher reliability. In order to enable an appropriate trade-off decision to be made between reliability and the required processing power, benchmarking of four CD algorithms has taken place on bases of a generic, culture-diverse multi-genre AV corpus. In terms of complexity/performance trade-off, a field-difference-based CD proved to be optimal.

Dense Structure-from-Motion: An Approach Based on Segment Matching

For 3D video applications, dense depth maps are required. We present a segment-based structure-from-motion technique. After image segmentation, we estimate the motion of each segment. With knowledge of the camera motion, this can be translated into depth. The optimal depth is found by minimizing a suitable error norm, which can handle occlusions as well. This method combines the advantages of motion estimation on the one hand, and structure-from-motion algorithms on the other hand. The resulting depth maps are pixel-accurate due to the segmentation, and have a high accuracy: depth differences corresponding to motion differences of 1/8/sup th/ of a pixel can be recovered

Temporal stabilization of video object segmentation for 3D-TV applications

We present a method for improving the temporal stability of video object segmentation algorithms for 3D-TV applications. First, two quantitative measures to evaluate temporal stability without ground-truth are presented. Then, a pseudo-3D curve evolution method, which spatio-temporally stabilizes the estimated object segments is introduced. Temporal stability is achieved by re-distributing existing object segmentation errors such that they are less disturbing when the scene is rendered and viewed in 3D. Our starting point is the hypothesis that if making segmentation errors are inevitable, they should be made in a temporally consistent way for 3D TV applications. This hypothesis is supported by the experiments, which show that there is significant improvement in segmentation quality both in terms of the objective quantitative measures and in terms of the viewing comfort in subjective perceptual tests. This shows that it is possible to increase the object segmentation quality without increasing the actual segmentation accuracy.

Computation of Green's function of laterally varying media by means of a modal expansion

In many cases the strongest signals in seismic data arise from scattering at shallow heterogeneities. An important step in modeling and removing this type of deterministic noise is computing Greens function for a 3D laterally varying model. To this extent an efficient method is developed, which is based on the combination of guided-mode expansions and ray-tracing techniques. This method is illustrated for a two layer model. I n t r o d u c t i o n The regions of highest interest in seismic exploration are the deeper ones. However, the strongest signals in seismic data generally arise from heterogeneities in the near-surface regions. These heterogeneities lead to scattering and to waves which contain no information on the deeper regions (e.g. head waves and Rayleigh waves). Many methods have been developed on an ad-hoc basis for removing these effects. Most of these methods are adaptive, and despite the fact that they have been succesful in many cases, they are not always able to deal with the complexity of the scattering process in the shallow subsurface. Examples of these methods are “filtering” , “statics” and “gapped deconvolution” (see for example Yilmaz, 1987). Recently, a method has been developed for removing the effects of heterogeneities in the shallow subsurface by means of linearized inversion and high-frequency asymptotics for the case of a laterally homogeneous model of the earth. This method has been applied succesfully to field data (see Blonk et al., 1994, 1995). The method which is described here is an extension of the former method for the case where the assumption of lateral invariance is no longer valid. Our method is also based on linearized inverse scattering and high frequency asymptotics. The method consists of the following steps: First an estimate of certain characteristics of the background model has to be made. We assume that the background model consists of (possibly heterogeneous) horizontal layers (see figure 1). Due to the layered structure of the background model, the propagation velocity of seismic waves in this background model is frequency-dependent. For this part of the method, concepts form integrated optics are used. The next part of the method consists of estimating the distribution of heterogeneities in the shallow subsurface. This has to be done by using as many data as possible. In this estimation procedure, Green’s function of the background model plays a vital role. The objective of this estimation procedure is to find a distribution which can explain the observed scattering. Due to the ill-posedness of the problem, this does not need to be the actual distribution. The last step in the method consists of determining the scattered field for each shot from the estimated scatterer distribution, and subtracting this field from the original data. In this paper, we concentrate on the computation of the headwave part of Greens function and restrict ourselves to the scalar case. Computa t ion of Green’s func t ion The general structure of the background model we consider here is depicted in figure 1. We have N layers, where for layer i we have < < and = 0. The Nth layer is infinitely thick. The governing equation for Green’s function in the frequency domain is given by:

From spatiotemporal curves to reconstructed depth

We present a novel approach for 3D reconstruction based on a set of images taken from a static scene. Our solution is inspired by the spatiotemporal analysis of video sequences. The method is based on a best fitting scheme for spatiotemporal curves that allows us to compute 3D world coordinates of points within the scene. As opposed to a large number of current methods, our technique deals with random camera movements in a transparent way, and even performs better in these cases than with restrained motion such as pure translation. Robustness against occlusion and aliasing is inherent to the method as well.