A knowledge based approach to finding fixed size rocks within a n image is described. It is intended t o be a building block within a multiresolution system. Each point is hypothesised t o be a rock and a region surrounding it is labelled using knowledge of rock characteristics. Twelve features are then measured and used t o test the hypothesis by means of a combination of thresholding and k-nearestneighbour classification.
I IntroductionMeasurement of rock size distribution or fragmentation has application in the mining industry where sophisticated control systems are used to monitor and control autogenous mills. T h e use of image processing as a non-invasive measurement technique offers advantages over conventional techniques such as: undisturbed feed monitoring and repeatble results. Hunter et a1 [l] gives an overview of image processing techniques for measuring fragmentation. See McDermott and Miles [a], Ivanov et a1 [3] and Berger [4] for further examples.This paper describes a knowledge based approach to determining the position of fixed size rocks within an image. This solution is being used as a building block within a larger multiresolution system for measuring fragmentation using image processing. To take advantage of the multiresolution approach, all processing has been matched t o a particular size of object, later referred t o as the optimal size. T h e radius of an optimally sized disc will be referred t o as the optimal radius. For a similar approach using neural networks see Crida [5]. T h e system described here will attemp to detect rocks at each level in a multiresolution image pyramid and compile a list of detected rocks from which the size distribution can be calculated.Although time consuming, this approach to the problem is more robust and accurate than more direct techniques which attempt t o measure a size distribution directly by means of chord measurements [B]. In addition the performance is more easily verified since each rock identified by the system can b e selected and highlighted on a n image.T h e process of detecting optimally sized rock will be referred to as the system. T h e system described here uses knowledge of rock images t o postulate rules which are appropriate when an optimally sized rock is present in the region of interest. In the first part of the system, it is initially hypothesised that each point in the image is the centre of an optimally sized rock. T h e rules are then used to determine the extent of a blob in a region of interest around each point by locking onto image features which could correspond t o an optimally sized rock. A blob is defined as a bright region within the grey level image, the extent of which is the segmented region. The first part of the system is broken into three stages; blob edge detection, boundary completion and blob extent calculation. These are described in Sections I1 and 111.The next stage involves testing the hypothesis. A feature vector is calculated for the detected blob extents and classification is performed t o decide if t...
