COGOLIN

Context

The COGOLIN is an operational and innovating machine vision system that provides the size and the morphology of bulk rock fragments scrolling on a conveyor belt in harsh industrial environments such as quarries or mines. It improves the continuous production process providing these measures in real real-time. It has been conceived and implemented for the mineral industry in the framework of a sustainable development.

The COGOLIN system

 

[image]

The IP67 COGOLIN prototype

The COGOLIN’s innovation is the use of 3D surface measurements. The scene is illuminated with a laser sheet of light. This creates a reflected line as seen from the light source, and appears as a curve from any other point out of the plane of the sheet. By observing the reflected sheet of light using high resolution camera and knowing the positions and orientations of both camera and light source, it is possible to determine the distance between the reflected points and the light source or the camera. By moving the scene (conveyor belt) a sequence of depth profiles of the scene can be generated and it can be represented as a range image. This on-the-edge technology allows us to acquire surface 3D images at high speed and thus keeping up with the industrial conveyor belt velocity.

A 2-years industrial research allowed to develop a perfectly optimized machine vision system (camera, lens and laser parameters combined) to obtain the best 3D images in terms of quality and spatial resolution. Calibrated COGOLIN images are free of distortion normally due to lens properties and perspective. This allowed us to measure directly real features in the image.

Why do you need COGOLIN?

The implemented concurrent feedback the COGOLIN Vision Machine helps you directly by non-intrusive and non-disruptive machine vision to:

  • improveyour quality control of the product,
  • optimizeyour production facilities,
  • reduceyour overall energy consumption.

What is the ideal client profile?

If you need a machine vision in harsh environment to control your exploitation operations and to control your costs and product quality, the COGOLIN Machine VIsion offers you the best solution.

Examples of business where COGOLIN is essential:

·        quarries environment,

·        underground mining,

·        aggregate industry,

·        cement industry,

·        blast optimization,

·        production lines,

·        tracking fragmentation,

COGOLIN Specifications

  • analyse the conveyor belt running up to 3,2 meters/second,
  • take up to 30.000 profiles/second at 1536 pixels width,
  • take up to 7.000 profiles/second at 4096 pixels width,
  • accomodate a high resolution specialised camera device from 0,8 to 12,6 Mpixels,
  • get an accurate spatial resolution of 0,92mm/pixel at high speed conveyor belt (i.e. 3,2m/s).

Where a typical setup is:

  • 60cm of width at 1536 pixels (0,46mm/pixel),
  • 11,7cm of hight at 256 pixels (0,92mm/pixel),
  • 3,2m/s speed of a conveyor belt,
  • 3000 images/second shot.

[image]

A 3D surface image triangulated using the laser sheet of light technique

Results

[image]

Graphics of the real-time size distribution by weight and the volume flow rate rate

Comparison between a 2D and a 3D systems

Property/Process 2D COGOLIN 3D
Segmentation Sensitive to lighting, contrast, camera exposure, historgram dynamic. Insensitive to lighting, contrast, camera exposure, histogram dynamic.
Calibration Sensitive to the tilt distortion and sensitive to the calibrated reference plane.

Calibration can only be performed on a reference plane. Fragments over/below the reference place are over/under-estimated.

Since the acquisition is a measure, rigorous calibration can be performed.

Exact calibration is applied allowing for 3D measurements regardless of the fragment position.

Size The size estimation is biased by the 2D calibration. While this bias is not important for monolayer analysis, it could become significant when the pile is high.

The size is estimated based on the equivalent disc diameter (based on fragment area projection).

The size measurement is not biased by the fragment position on the conveyor belt.

The size is measured by the maximum inscribed disc (correlated to the sieve diamater).

COGOLIN provides other size parameters such as the width, the length and the depth. All these parameters have physical meaning and they are used to compute shape parameters like the elongation and flakiness index (as defined in Norm NFP 18561 and BS812)

Shape analysis Only the elongation could be computed when the fragments are in a monolayer. Elongation and flakiness index are computer accurately when fragments are in a monolayer. Those parameters can be computed in pile but with a lower precision.
Volume Individual fragment volume estimation is based on a hypothesis of a spherical model of the fragments. It implies bias for flat aggregates. Individual fragment volume measurement is computed from the 3D upper surface of the fragment and based on the hypothesis that the fragment is symmetric with respect to an interpolated mid-plane. Bias has to be considered when fragments are very irregular (high concavity).
Size distribution The size distribution by mass suffers from the 2D size and volume limitations. It is commonly biased when the pile is high or when the fragments are flat. The size distribution by mass is more robust and more accurate but it could suffer bias when the fragments are very irregular (high concavity for instance).

Comparison between a 2D technique (WipFrag, Split, FragScan, Fragalyst, …) and a 3D technique (COGOLIN)