Flyrock produced by blasts in surface mining and quarrying is a major hazard that can result in serious injuries or property damage. This paper presents a fully integrated and automated approach for reviewing blast outcomes from a wide variety of rock blasting environments from the perspective of flyrock safety associated with blasthole collar ejections. The process is applied at a corporate level for an explosives service provider handling up to 100 blasts per day from mines, quarries and construction sites throughout the US.
Flyrock risk minimisation has inspired research and the associated development of a reliable flyrock model to estimate flyrock ejection velocities and the resulting maximum projection distances of fragments. In spite of significant achievements in this area, flyrock incidents continue to occur, mostly due to implementation mistakes. An efficient preventive measure is continuous blasting process improvement that relies on on-site video recording, storage, automatic analysis and classification of the recorded blasts followed by the automatic generation of reports.
In this paper, a novel blasting video analysis system that is capable of processing a large quantity of video records is presented. Blast analysis and parameter measurement are fully automated and are carried out using videos recorded with a conventional consumer camera. The blast video recording process does not require the use of surface markers, range poles or objects of a predefined size for scaling purposes.
Although automatic video analysis is commonly used in many application areas, such as video surveillance, traffic monitoring and robotics, the algorithms that have been developed to recognise and track solid objects with well-defined shapes, textures and colours are not efficient at dealing with blasts, which, by their nature, have dynamically changing parameters. Moreover, very strong camera vibration caused by blasts that cannot be totally compensated by camera image stabilisers may cause conventional computer vision algorithms to fail.
This paper presents novel computer vision methods developed for automatic tracking of the trajectory from a blast ejection point to the highest point reached by blast particles and the ejection velocity. The system also uses colour and motion analysis to detect the emission of toxic fumes during a blast. The solution is vibration tolerant and can differentiate foreground and background data, even in the presence of camera vibration. Moreover, the blasting video analysis method can automatically categorise correctness of the measurements and therefore indicate cases when additional visual inspections may be required to verify reports generated automatically.
Kharitonenko, I, McKenzie, C K, Papillon, B E and Popov, P, 2015. An automated system for flyrock and fume monitoring of blasts, in Proceedings 11th International Symposium on Rock Fragmentation by Blasting, pp 417–424 (The Australasian Institute of Mining and Metallurgy: Melbourne).