Fibre Measurement: Automated Dimensional Analysis of Industrial Reinforcing Fibres

This page describes our approach to a technically demanding metrology application: automatically measuring the length and diameter of large quantities of randomly presented industrial fibres, where the accuracy and distribution of measurements must satisfy defined quality specifications.

The Application

A manufacturer of reinforced materials is required to demonstrate that the fibres used in their product meet defined dimensional specifications — specifically the distribution of fibre lengths and diameters across each production batch. To satisfy this requirement, a sample of fibres from each batch is processed through a measurement system and the resulting distribution is compared against the specified thresholds.

Previously this process required significant manual effort and operator attention. The client required an automated solution capable of processing large fibre samples accurately and with minimal operator involvement.

The Challenge

The fibres presented several significant technical challenges:

Fibre geometry — the fibres are never straight. They are bent, curved, hooked, and irregular in shape. Standard length measurement approaches — which assume a broadly linear object — are not applicable. A more sophisticated approach was required to accurately trace and measure each fibre along its actual path.

Touching and crossing fibres — in a natural, uncontrolled presentation, fibres inevitably touch and cross each other. The ability to separate and measure individual fibres even when in contact with neighbours was critical — without this capability, sample preparation would require fibres to be manually separated before measurement, adding significant time and effort to the process.

Variable diameter — fibre diameter is not uniform along the length of each fibre. A meaningful measurement requires diameter to be sampled at multiple points along the fibre, with minimum, maximum, and average values extracted.

Throughput — the system needed to process an entire batch sample continuously and without constant operator attention, updating results in real time until the full sample had been measured.

The Solution

Imaging setup

Fibres are dispensed onto a semi-transparent conveyor belt, with a collimated backlight positioned beneath. A 12-megapixel camera with a high-resolution lens is mounted above the conveyor, with image acquisition linked to conveyor speed. The frame rate is adjusted dynamically with conveyor speed to ensure that the longest fibre in the sample is captured in its entirety within at least one frame.

Handling repeated imaging

Because smaller fibres may pass through the field of view across multiple acquisitions, the system tracks conveyor travel between frames. Fibres that have already been measured are identified and excluded from subsequent processing — ensuring each fibre is measured exactly once regardless of its size relative to the field of view.

Fibre segmentation

Each fibre is segmented from its neighbours using image processing techniques designed to handle contact and crossing between fibres. This is one of the most technically demanding aspects of the application — reliably separating individual fibres even when they touch, cross, or partially overlap, without requiring manual preparation of the sample.

Length measurement

Once segmented, each fibre is skeletonised — its perimeter is reduced to a single-pixel-wide centreline that traces the fibre's actual path, including all curves, bends, and hooks. The length of this skeleton provides an accurate measurement of the true fibre length regardless of geometry.

Diameter measurement

Diameter is measured at equal intervals along the length of each fibre. Minimum, maximum, and average diameter values are extracted for each fibre and recorded.

[IMAGES]

Outcome

The system processes entire batch samples continuously and without operator intervention, updating a real-time display showing the fibre length and diameter distribution curves and whether the batch meets the defined specification thresholds. The ability to measure touching and crossing fibres without manual separation significantly reduced the preparation effort previously required, allowing large batch samples to be processed accurately and efficiently.

The system was commissioned successfully and is operating in production.

Working with Oculus Vision

If you are specifying or delivering a project involving dimensional measurement of complex or irregular objects, high-volume automated metrology, or inspection where object separation and segmentation present a significant technical challenge, we are happy to discuss the application and whether a similar approach would be appropriate.