Skip to main content
European Commission logo print header

DEVELOPMENT AND TESTING OF NEW TECHNIQUES FOR MINERAL EXPLORATION BASED ON REMOTE SENSING, IMAGE PROCESSING METHODS AND MULTIVARIATE ANALYSIS.

Obiettivo

DEVELOPMENT OF AN OPERATIONAL SYSTEM FOR EXPLORATION USING MULTIDATA CORRELATION INCLUDING REMOTE SENSING IMAGERIES.
New software was developed for statistical image processing, treatment of structural data, and data integration. It was considered important to adapt the software developed for use on microcomputers. Research involved:
integration of datasets;
automatic detection of lineaments;
classification and multivariate analysis of the datasets;
development of 3-dimensional models;
software development.

In the analysis of large amounts of geochemical samples calibration problems occur sometimes giving chessboard patterns in maps showing the analyses of the different elements. The method developed for correcting these patterns seem to yield very good results.

The multivariate analyses (factor analysis, minimum/maximum autocorrelation factors (MAF), discriminant analysis) served a 2-fold purpose. One was acting as a quality control measure, the other to give further insight in the distribution of element concentrations.

The prediction of heavy mineral counts by means of a classification and regression tree (CART) procedure was found to be very useful. For different heavy minerals, areas with a high potential have been delineated, and the predictions have been confirmed by other investigations.

Methods like 'the grand tour' and projection pursuit have potential within datasets with many channels (imaging radiometers) and the feature selection algorithm based on Jeffreys-Matusitas distance has already shown a superior behaviour in many applications.

One of the major tasks in structural analysis of geological image data is the mapping of lineaments and the subsequent computation of lineament statistics. Based on microrelief properties, a prototype procedure for estimating the distribution of the direction of lineaments was further developed incorporating an actual lineament drawing procedure.

The estimation of 3-dimensional elevation models based on stereo pairs of images by means of direct matching (as opposed to feature matching) was evaluated by a visual c omparison between the estimated height values and the topographic maps of the area. The result of this comparison was very encouraging.
IN THE FRAMEWORK OF THE ALMADEN JOINT PROJECT, IMSOR WILL BE RESPONSIBLE FOR THE INTEGRATION OF DATA SETS, AUTOMATIC DETECTION OF LINEAMENTS, CLASSIFICATION AND MULTIVARIATE ANALYSIS OF THE DATA SETS.

THE COMBINATION OF GEOPHYSICAL, GEOCHEMICAL AND MULTISPECTRAL DATA IN A COMMON DATA BASE INVOLVES RESAMPLING TO A UTM GRID, GEOSTATICAL MODELLING AND INTERPOLATIONS B.M.O. KRIGING TECHNIQUES AND BICUBIC SPLINE FITS. THE MULTIVARIATE ANALYSIS AND CORRELATION OF COMBINED DATA SETS INVOLVES ORTHOGONAL TRANSFORMATIONS LIKE PRINCIPAL COMPONENTS, FACTOR ANALYSIS, AND CANONICAL CORRELATION ANALYSIS. THE SOCALLED MAF FACTORS TAKE THE SPATIAL CORRELATION INTO ACCOUNT IN THE COMPUTATION OF ORTHOGONAL FACTORS AND THEIR APPLICABILITY SHOULD BE FURTHER EVALUATED. THE AUTOMATED ESTIMATION OF LINEAR FEATURES WILL BE DEVELOPED USING TEXTURE STATISTICS, DATA DEPENDENT FILTERS AND OTHER DETERMINATIONS OF MICRORELIEF. THIS WILL INVOLVE A HIERARCHICAL DATA REPRESENTATION. FURTHERMORE DISCRIMINATION PROCEDURES USING CONTEXTUAL INFORMATION WILL BE DEVELOPED AND TESTED. THESE SHOULD PARTLY INVOLVE MODELS FOR THE SPATIAL CONTINUITY OF THE POPULATIONS AND ON FEATURE VECTORS DESCRIBING LOCAL TEXTURAL FEATURES. FINALLY THERE SHOULD BE DEVELOPED 3D MODELS (BASED ON SPOT IMAGERY) IN ORDER TO EVALUATE THE TOPOGRAPHIC INFLUENCE ON THE MULTISPECTRAL IMAGERY.

Argomento(i)

Data not available

Invito a presentare proposte

Data not available

Meccanismo di finanziamento

CSC - Cost-sharing contracts

Coordinatore

Danmarks Tekniske Universitet
Contributo UE
Nessun dato
Indirizzo

2800 Lyngby
Danimarca

Mostra sulla mappa

Costo totale
Nessun dato

Partecipanti (3)