No abstract

Properties of objects and spatial relations between objects play an important role in rulebased approaches for high-level vision. The partial presence or absence of such properties and relationships can supply both positive and negative evidence for region labeling hypotheses. Similarly, fuzzy labeling of a region can generate new hypotheses pertaining to the properties of the region, its relation to the neighboring regions, and finally, the labels of the neighboring regions. In this paper, we present a unified methodology to characterize properties and spatial relationships of object regions in a digital image. The proposed methods can be used to arrive at more meaningful decisions about the contents of the scene.

No abstract

Mapping fault planes using seismic images is a crucial and time-consuming step in hydrocarbon prospecting. Conventionally, this requires significant manual efforts that normally go through several iterations to optimize how the different fault planes connect with each other. Many techniques have been developed to automate this process, such as seismic coherence estimation, edge detection, and ant-tracking, to name a few. However, these techniques do not take advantage of the valuable experience accumulated by the interpreters. We have developed a method that uses the convolutional neural network (CNN) to automatically detect and map fault zones using 3D seismic images in a similar fashion to the way done by interpreters. This new technique is implemented in two steps: training and prediction. In the training step, a CNN model is trained with annotated seismic image cubes of field data, where every point in the seismic image is labeled as fault or nonfault. In the prediction step, the trained model is applied to compute fault probabilities at every location in other seismic image cubes. Unlike reported methods in the literature, our technique does not require precomputed attributes to predict the faults. We verified our approach on the synthetic and field data sets. We clearly determined that the CNN-computed fault probability outperformed that obtained using the coherence technique in terms of exhibiting clearer discontinuities. With the capability of emulating human experience and evolving through training using new field data sets, deep-learning tools manifest huge potential in automating and advancing seismic fault mapping.

No abstract

No abstract

Many previously published wave-equation-based methods, which attempt to automatically invert traveltime or kinematic information in seismic data or migrated gathers for smooth velocities, suffer a common and severe problem — the inversions are involuntarily and unconsciously hijacked by amplitude information. To overcome this problem, we have developed a new wave-equation-based traveltime inversion methodology, referred to as full-traveltime (i.e., fully dependent on traveltime) inversion (FTI), to automatically estimate a kinematically accurate velocity model from seismic data. The key idea of FTI is to make the inversion fully dependent on traveltime information, and thus prevent amplitude interference during inversion. Under the assumption that velocity perturbations cause only traveltime changes, we have derived the FTI method in the data and image domains, which are applicable to transmitted arrivals and reflected waves, respectively. FTI does not require an accurate initial velocity model or low-frequency seismic data. Synthetic and field data tests demonstrate that FTI produces satisfactory inversion results, even when using constant velocity models as initials.

No abstract