An Autonomous Vision-based Approach for On-shovel Rock Size Estimation in Open Pit Mining Industry

Rock Engineering 2009 - Rock Engineering in Difficult Conditions
Nima Ziraknejad, Hairong Zeng, Edmond Chow, Arya Ohadi, Shahram Tafazoli,
Abstract Estimating the size of rocks after blasting operation is an important process in today’s advanced mining industry. Blast engineers require rock size related statistics to carefully adjust the amount of blasting materials used in different sections of a mine. Examples of such statistics are the so-called P numbers (e.g. P80 and P100). The rock size requirements vary depending on ore type and crusher specifications. In this paper an autonomous on-shovel bucket-based rock fragmentation and size estimation method using a camera installed on top of the boom structure of electric and hydraulic mining shovels is presented. A securely installed heavy-duty camera provides the required imaging data to an industrial embedded computer system located inside the cab. The camera provides a series of bucket images with a sampling rate of 30Hz. A real-time machine vision software continuously analyzes the captured images and selects the most suitable bucket image (full of rocks and close enough to the camera) in every digging/dumping cycle. It then passes the obtained image to the rock segmentation routine running on the same computer platform. Intensive image processing procedures are applied to the obtained suitable bucket image in order to provide an enhanced image to the developed rock segmentation algorithm. The sizes of the segmented rocks in the image frame (640x480) are transformed to their real-world sizes using a linear calibration method.
Keywords: Rock fragmentation, Rock size estimation
Full Access to Technical Paper
PDF version for $20.00
Other papers from Rock Engineering 2009 - Rock Engineering in Difficult Conditions