Daniel Wedge

Trajectory Based Video Sequence Synchronization

Video sequence synchronization is often necessary for computer vision applications where multiple simultaneously recorded videos are processed. We present a coarse-to-fine approach to synchronizing two video sequences recorded at the same frame rate by stationary cameras with fixed internal parameters. At the coarse level, each sequence is broken into a set of sub-sequences, which are then matched. A voting scheme determines the range in which the sequences’ temporal offset lies. The fine synchronization step searches for the temporal offset by initially examining integer offsets, and then using the golden-section search to locate the offset to sub-frame accuracy. Our algorithm recovers the temporal offset of two sequences using the motion of a single moving object, is computationally efficient, and it does not require any stationary background points as reference points. We present results for synthetic data and real video sequences, with various degrees of temporal overlap.

Motion guided video sequence synchronization

We present an algorithm that synchronizes two short video sequences where an object undergoes ballistic motion against stationary scene points. The object’s motion and epipolar geometry are exploited to guide the algorithm to the correct synchronization in an iterative manner. Our algorithm accurately synchronizes videos recorded at different frame rates, and takes few iterations to converge to sub-frame accuracy. We use synthetic data to analyze our algorithm’s accuracy under the influence of noise. We demonstrate that it accurately synchronizes real video sequences, and evaluate its performance against manual synchronization.

Evaluation of automated lithology classification architectures using highly-sampled wireline logs for coal exploration

Wireline logs are a supplemental data source to conventional core logging. The recent explosion of machine learning algorithms has provided researchers with the opportunity to develop advanced statistical tools for automatically classifying lithology from these logs, enabling geologists to rapidly produce first-pass interpretations and validate others, even when core samples are missing or damaged. However, the machine learning algorithms need to be evaluated in the case where wells contain a large number of wireline logs which are highly-sampled. This paper explores different machine learning algorithms and architectures for classifying lithologies (e.g. coal and sandstone) using wireline data from a project area well-known for coal mineralisation: Juandah East, 60km north-west of Wandoan (Queensland, Australia). We used data from seven wells, each containing 19 wireline logs uniformly sampled at 1cm, retrieved from the open Queensland Digital Exploration (QDEX) database.Three popular supervised machine learners, namely the Naive Bayes classifier, Support Vector Machine, and Artificial Neural Network, were tested under two architectures: committee (one classifier per well log) and singular (one classifier for all well logs). Favourable performance was achieved under both architectures when the base classifier was tuned to maximise a coal-specific performance metric. Results show that the committee architecture increased overall accuracy, generally by increasing accuracy on the dominant lithology class and reducing the classification rate of minor lithology classes. Overall accuracy was further improved by post-processing to remove thin classified intervals ( < 10 cm ). The committee architecture provides the benefits of faster classifier training time through parallelisation, as well as a flexible platform for incorporating additional well logs without the need to retrain existing classifiers. HighlightsMachine learning methods classify lithologies using highly sampled wireline logs.Different classifiers and architectures are compared for practical industry use.A committee architecture provides parallelised well-based lithology classification.Lithology classifiers are optimised for coal exploration.

An irregular triangle mesh buffer analysis method for boundary representation geological object in three-dimension

Three-dimensional (3D) buffer analysis is a basic function of spatial analysis used widely in 3D Geographic Information Systems (3DGIS). Current buffer analysis methods for spatial points and curves generally function well. One exception is buffer zone of surface. Previous researchers in this field have used voxel models to overcome this limitation; however, defects with voxel model buffer analysis include redundancies, approximations, and poor visualization characteristics. In this contribution, a surface buffer analysis method is presented for the boundary representation of geological objects. Exact geometric representation is achieved via the construction of irregular triangle meshes in 3D. The results can be used for 3D structural modeling and then form the basis for spatial analysis or model-based quantitative assessment in mineral potential mapping and resource evaluation. Three comparisons between existing voxel methods and our new method, evaluating visualization, precision and redundancy, are conducted. The comparisons show that our proposed method is robust and provides a higher quality output than voxel modeling. Finally, uncertainty analysis of buffer distance in different geological objects was discussed.

Improving assessment of geological structure interpretation of magnetic data

Geological structures are recognisable as discontinuities within magnetic geophysical surveys, typically as linear features. However, their interpretation is a challenging task in a dataset with abundant complex geophysical signatures representing subsurface geology, leading to significant variations in interpretation outcomes amongst, and within, individual interpreters. Previously, numerous computational methods were developed to enhance and delineate lineaments as indicators for geological structures. While these methods provide rapid and objective analysis, selection and geological classification of the detected lineaments for structure mapping is in the hands of interpreters through a time consuming process. This paper presents new ways of assisting magnetic data interpretation, with a specific aim to improve the confidence of structural interpretation through feature evidence provided by automated lineament detection. The proposed methods produce quantitative measures of feature evidence on interpreted structures and interactive visualisation to quickly assess and modify structural mapping. Automated lineament detection algorithms find the feature strengths of ridges, valleys and edges within data by analysing their local frequencies. Ridges and valleys are positive and negative line-like features detected by the phase symmetry algorithm which finds locations where local frequency components are at their extremum, the most symmetric point in their cycle. Edge features are detected by the phase congruency algorithm which finds locations where local frequency components are in phase. Their outputs are used as feature evidence through interactive visualisation to drive data evidenced interpretation.Our experiment uses magnetic data and structural interpretation from the west Kimberley region in northern Western Australia to demonstrate the use of automated analysis outputs to provide: quantitative measures of data evidence on interpreted structures, and graphical evaluation of interpretation quality.

Constraining gravity gradient inversion with a source depth volume

Efficiently extracting the maximum amount of information from gravity gradient data is challenging. Interpretation often takes place in either the data domain or model domain. Here, we present a workflow that utilizes two interpretation techniques that can result in better characterization of the subsurface. Using a method that estimates depth to source, we obtain a depth volume of estimated source locations. The depth volume is then used to constrain inversion of gravity gradient data in the form of a reference model and 3D model weighting. We demonstrate that this combined approach improves the ability to recover sources at depth.

Learning characteristic natural gamma shale marker signatures in iron ore deposits

Abstract Uncertainty in the location of stratigraphic boundaries in stratiform deposits has a direct impact on the uncertainty of resource estimates. The interpretation of stratigraphic boundaries in banded iron formation (BIF)-hosted deposits in the Hamersley province of Western Australia is made by recognizing shale markers which have characteristic signatures from natural gamma wireline logs. This paper presents a novel application of a probabilistic sequential model, named a continuous profile model, which is capable of jointly modelling the uncertainty in the amplitude and alignment of characteristic signatures. We demonstrate the accuracy of this approach by comparing three models that incorporate varying intensities of distortion and alignment in their ability to correctly identify a shale band of the West Angelas member of the Wittenoom Formation which overlies the Marra Mamba Iron Formation in the Hamersley Basin. Our experiments show that the proposed approach recovers 98.72% of interpreted shale band intervals and importantly quantifies the uncertainty in scale and alignment that contribute to probabilistic interpretations of stratigraphic boundaries.

Evaluation of Automated Lithology Classification Architectures using Highly-Sampled Wireline Logs for Coal Exploration

Wireline logs are a supplemental data source to conventional core logging. The recent explosion of machine learning algorithms has provided researchers with ample opportunity to develop automated statistical tools for classifying lithology from wireline logs, which geologists can use to produce first-pass interpretations or to validate existing interpretations. Such automated interpretations can be particularly valuable information in the case of missing or damaged core samples. There exists, however, a need to evaluate said machine learning algorithms in the case where available wireline logs contain a wide range of different logs which are highly-sampled. This paper explores different machine learning algorithms and architectures for lithology classification using wireline data from project area Jundah East, 60 km north-west of Wandoan, Queensland, which is well known for coal mineralisation. We used seven well logs each containing 19 wireline logs sampled at 1 cm-1, available through the Queensland Digital Exploration (QDEX) data system. Three popular supervised machine learners, namely the Naive Bayes classifier, Support Vector Machine, and Multilayer Perception (an artificial neural network), are tested under two architectures: committee (one classifier per well log) and singular (one classifier for all well logs). The results show the Naive Bayes classifier, although computationally simple, achieves good results in general when training using a committee architecture on a large data set. For coal classification in particular, it achieved the sensitivity score of 0.79 and the specificity score of 0.97. While the committee and singular architectures generated similar results, the committee architecture provides the benefits of faster computation time as well as a flexible platform for the training of additional well logs.

Automated Structure Detection and Analysis in Televiewer Images

Borehole televiewer data is an important source of data on structural and stratigraphic discontinuities in both the mining and petroleum industries. Manually picking features in downhole image logs is a labour-intensive and hence expensive task and as such is a significant bottleneck in data processing. It is also a subjective process. We present a new algorithm and workflow for automatically detecting and analysing planar structures in downhole acoustic and optical televiewer images. First, an image complexity measure highlights areas most suitable for automated structure detection. Changes in the image complexity can be used to locate geological boundaries. Second, structures are automatically detected, with each structure having an associated confidence level; users can apply a threshold to the confidence values to adjust the quality and quantity of the detected structures based on the image quality and geological complexity. Third, structures that have been detected but that do not meet the structure confidence threshold can be interactively assessed and if necessary selected. We also provide tools for rapidly picking sets of equivalent structures and reducing structures to a set of representative picks.