A Low-Complexity Algorithm for Estimation of The P80 Rock Granulometry Indicator Based on Digital Image Processing

This work presents an algorithm that automatically measures rock size distribution during primary crushing in mining, continuously monitoring without interrupting production or causing issues in subsequent crushing stages due to oversized rocks. Existing commercial solutions demand significant investment and infrastructure, making them unfeasible for small mining operations. As a result, these companies typically halt production and send samples to a laboratory to determine the P80 granulometric factor. Most current research focuses on edge detection and rock segmentation, often overlooking the estimation of granulometric parameters. The proposed algorithm uses low-complexity image processing techniques that can run on a small-board computer. Images are captured by a camera positioned above a conveyor belt within an image acquisition enclosure and processed using watershed segmentation and morphological operations like erosion and dilation. The P80 value is estimated through the Rosin-Rammler linearization model. Results show 89% accuracy compared to laboratory measurements and a 91.3% success rate.