Objective
IDENTIFICATION OF MINERAL FIBROUS PARTICLES AND CLASSIFICATION ACCORDING TO THEIR DIMENSIONS AND TYPE.
THE RESEARCH PROJECT IS IMPLEMENTED IN TWO PHASES:
1. PRE-PROCESSING
THE IDENTIFICATION OF FIBRES IS FEASIBLE AFTER ADEQUATE PRE-PROCESSING. COMPUTING CAPABILITY HAS BEEN EXTENDED BOTH IN QUALITY AND SPEED SINCE THE COMPLETION OF THE DELFT IMAGE PROCESSOR (DIP), A COMPUTED SYSTEM SPECIALLY DESIGNED AND BUILT FOR FAST LINEAR AND NON-LINEAR FILTERING. THIS SYSTEM ALLOWS OPERATIONS LIKE DELINERATION OF THE PARTICLES (SEGMENTATION), ETCHING (EROSION) AND THICKENING (DILATION) OF THEIR SILHOUETTES.
2. DECOMPOSITION OF COMPLEX STRUCTURES
THE TYPICAL FIBRE SHAPE CAN BE EXPLOITED THROUGH THE HOUGH TRANSFORM TECHNIQUE, WHICH IS MAINLY SENSITIVE TO LONG STRAIGHT STRUCTURES. IN ORDER TO IMPROVE SELECTIVITY, THE ORIENTATION OF THE PARTICLE BORDERS IS TAKEN INTO ACCOUNT AS A WEIGHTING FACTOR.
THIS METHOD CAN BE COMBINED WITH THE LINE-TRACKING METHOD, IN WHICH A FIBRE IS ROUGHLY FOLLOWED THROUGH THE IMAGE. DURING THIS PROCESS INFORMATION OF THE FIBRE BUILDS UP GRADUALLY AND ENABLES JUMPING ACROSS COMPLICATED AREAS.
Topic(s)
Data not availableCall for proposal
Data not availableFunding Scheme
CSC - Cost-sharing contractsCoordinator
2600 JA Delft
Netherlands