Three dimensional data analysis for the separation and sizing of laboratory rock piles
2002 (English)Doctoral thesis, comprehensive summary (Other academic)
The size distribution of piles of blasted and crushed rock in mining is so closely related to the economics of the mine that it must be measured quickly and accurately. This thesis develops a mathematical and 3D imaging solution for determining the size distribution of laboratory rock piles based on surface measurements of the pile.
Place, publisher, year, edition, pages
Monash University, 2002.
Research subject Signal Processing
IdentifiersURN: urn:nbn:se:ltu:diva-17323Local ID: 2d75db60-2374-11dd-ac76-000ea68e967bOAI: oai:DiVA.org:ltu-17323DiVA: diva2:990325
Upprättat; 2002; 20080516 (mjt); Examiner 1 Report prof.dr.ir. Lucas J. van Vliet It was a pleasure reading the aforementioned manuscript in which the candidate presents all aspects (not in equal depth) of predicting the size distribution of rock piles from surface observations using range images. A difficult real-world problem that is worth studying, not only because of its practical use, but also because its solutions have a much wider applicability. I am certainly impressed by the broadness of this work. The main apects are: 1. Development of sample preparation techniques to avoid bias in subsequent measurements. 2. Acquisition of range images using a dedicated home built system. 3. Fusion of range images that are acquired under multiple orientations resulting in a non-regular surface sampling grid. 4. Improvements and refinements of the Chavez surface probability model that proved critical to this research. 5. Elaborate testing of the improved surface probability model. 6. Development of image processing algorithms (morphology) that can work with irregularly sampled surface data. 7. Development of a segmentation method for range images to partition an image into entirely and partially visible rocks. 8. Hueristics based classification of surface rocks based on an automatic measure of the (visible) surface area and a visibility measure. I have added some remarks, comments and questions. I feel that the candidate can easily adopt these suggestions in his final thesis and I hope the Ph.D. committee will agree with me. ----------------- Examiner 2 Report Dr Hughues Talbot Summary In simple terms, this thesis essentially is about using the information present on the the top of a bucket full of rocks of various sizes and shapes to derive the rock size distribution present in the whole bucket. This problem has applications in mining. The data on the top of the rock pile is acquired in image form by structured lighting equipment and converted to irregularly spaced 3D data points corresponding to samples of the rock surface. This data is analysed automatically using mathematical morphology algorithms that have been extended to unstructured 3D points to provide information on the fully or partly visible individual rocks. The proportion of rocks in the whole pile is then derived using a surface probability model developed for this project. Experimental results are provided and calculated proportions are shown to agree with actual proportions within a few percentage points. Appreciation The problem Matthew has tackled is a difficult and interesting one, which is not limited to rocks but is in fact a fairly common problem. The literature review showed that no good solution existed prior to this research work. Overall, the work presented in this thesis is thorough and careful with regard to the experimentation, the image analysis of unstructured 3D data points is novel and interesting, the probabilistic approach to derive the proportion in the whole pile from the top layer is also sound and interesting. Finally the experimental results are indicative that the method has great potential. I am convinced that Matthew has displayed the skills necessary to conduct independent quality research and should be awarded his Ph.D.2016-09-292016-09-29