Introduction to Dynamic Image Analysis

1,909 views
1,763 views

Published on

Dr. Jeff Bodycomb from HORIBA Particle discusses the technology behind dynamic image analysis, advantages over sieves, and practical considerations.

This presentation is archived with the original webinar video in the Download Center at www.horiba.com/us/particle.

Published in: Technology
1 Comment
0 Likes
Statistics
Notes
  • Be the first to like this

No Downloads
Views
Total views
1,909
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
43
Comments
1
Likes
0
Embeds 0
No embeds

No notes for slide

Introduction to Dynamic Image Analysis

  1. 1. Introduction to Dynamic Image Analysis Jeffrey Bodycomb, Ph.D. HORIBA Scientific www.horiba.com/us/particle© 2010 HORIBA, Ltd. All rights reserved.
  2. 2. Why Image Analysis? Need shape information, for example due to importance of powder flow Verify/Supplement diffraction results Replace sieves for size distribution analysis Decide whether you have single particles or agglomerates Seeing is believing These may have the same size (cross section), but behave very differently.© 2010 HORIBA, Ltd. All rights reserved.
  3. 3. Why Image Analysis Need shape information for evaluating •Packing •Flow •Tendency to create dust •Abrasive performance •Optical properties (e.g., reflectivity)© 2010 HORIBA, Ltd. All rights reserved.
  4. 4. Why Dynamic Image Analysis? Robust measurement….the interaction between the instrument and the particle is optical, so there is no wear and change in calibration. High resolution size distribution results Fast Also, these are all reasons to use Dynamic Image Analysis instead of sieves.© 2010 HORIBA, Ltd. All rights reserved.
  5. 5. Sieves Weigh % sample caught on known screen sizes Solid particles 20 μm – 125 mm Low equipment cost Direct measurement method Some automation/calculation available More information available through www.retsch.com© 2010 HORIBA, Ltd. All rights reserved.
  6. 6. Sieves Tend to wear over time Difficult to tell when sieve results are “drifting” due to wear Results depend on nature of “shaking” leading to operator to operator variations in results. Rotap or Vibratory or Manual? Sieve overloading (too much material) intervals? How long do you shake? Amplitude? Limited information Finite number of sieves No shape information Cubes and needles interact in unexpected ways with the sieve.© 2010 HORIBA, Ltd. All rights reserved.
  7. 7. Two Approaches Static: Dynamic: particles fixed on slide, particles flow past camera stage moves slide 0.5 – 1000 um 2000 um w/1.25 objective 1 – 3000 um© 2010 HORIBA, Ltd. All rights reserved.
  8. 8. Major Steps in Image Analysis Image Acquisition and enhancement Object/Phase detection Measurements© 2010 HORIBA, Ltd. All rights reserved.
  9. 9. Dynamic Image Analysis: Moving Particles FeaturesUse gravity, or, better, vacuum (from a compressed air supply andventuri) in order to draw particles through instrument. Vacuumhelps keep the windows clean.© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 9
  10. 10. Acquiring Images We want a good microscope and nice sharp images. Pay attention to lighting and focus. No Yes© 2010 HORIBA, Ltd. All rights reserved.
  11. 11. More Light better Sharpness weaker light source => wider aperture blurry images = bad resolution wider depth of focal plane sharper images = more resolution stronger light source => smaller aperture => better images© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 11
  12. 12. Image Capture Speed (Time)CCD-Basicpixel raster Height of fallCCD-Zoom duringpixel raster 20 µm capturing time Slower (100 µs) Pixel = due to weak light source. Picture Element © 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 12
  13. 13. Capturing time influence CCD-Basic pixel raster CCD-Zoom Height pixel raster 15 µm of fall during illumination time Bright light source (and sensitive CCD) means no blurring.© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 13
  14. 14. Dispersing a Sample Want to spread particles out so that they don’t touch. No Yes© 2010 HORIBA, Ltd. All rights reserved.
  15. 15. Control feed rate. Want to spread particles out so that they don’t touch. Use % of field of view that is covered in order to control feed rate. Try 1% at first. Feeding Too fast Good© 2010 HORIBA, Ltd. All rights reserved.
  16. 16. Resolution Measuring Principle CCD - Basic CCD - Zoom© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 16
  17. 17. Two-Camera-System Measuring Principle m Zoo c Basi Basic-Camera Zoom-Camera© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 17
  18. 18. Resolution Measurement Results mixture of six sizes of grinding balls© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 18
  19. 19. Digital Image ProcessingArea Measurement Sieving Competing Measuring Methods A xarea A‘ = A xarea “diameter comparison via projection surface” CAMSIZER-measurement xarea (red) and sieving * (blue)© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 19
  20. 20. Digital Image ProcessingMeasuring of Width Sieving Competing Measuring Methods xc min Q3 0.9 --- width measurement “width” 0.8 -*- Sieving Tinovetin-B-CA584A_BZ_xc_min_002.rdf Syngenta-1mm-2min-Sieb.ref 0.7 0.6 0.5 xc min 0.4 0.3 0.2 0.1 0 0.1 1 x [mm] comparison CAMSIZER-measurement xc min (red) and sieving * (black)© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 20
  21. 21. CAMSIZER Sieving Competing Measuring Methods rice Q3 [%] 80 70 xc min 60 50 40 xarea 30 A A‘ = A 20 10 0 1.0 1.25 1.5 1.75 2.0 x [mm] xc min (width) is more similar to sieve result than xarea© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 21
  22. 22. Digital Imaging Sieving Competing Measuring Methods Q3 [%] Sample A_BZ_0.2%_xc_min_001.rdf Sample A_.ref 80 70 60 50 40 30 20 10 0 0.2 0.4 0.6 1 x [mm]© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 22
  23. 23. Fitting of CAMSIZER results to Sieving Competing Measuring Methods fitted result CAMSIZER-measurement xarea (red) to sieving * (blue)© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 23
  24. 24. Digital Imaging Sieving Competing Measuring Methods Q3 [%] 90 80 RT669_3993_Z_LB_05%_xc_min_001.rdf 70 RT669_RT_3993.ref 60 50 40 30 20 10 Examples of samples with 0 200 lens shaped particles 400 600 800 x [µm] without fitting CAMSIZER-result xc min (red) sieve analysis * (black)© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 24
  25. 25. Digital Imaging Sieving Competing Measuring Methods Use CAMSIZER data from a single sieve class (an “element” of the distribution” to provide superior fitting. q3 [1/mm] 0.7 New elementary 0.6 fitting with 0.5 0.4 0.3 single 0.2 (narrow) 0.1 sieve class Q3 0 1 2 3 4 xc_min [mm] 0.9 0.8 Two samples 0.7 0.6 0.5 with similar 0.4 0.3 0.2 shape 0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 b/l© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 25
  26. 26. Digital Imaging Sieving Competing Measuring Methods Q3 – Fitting does not fit both samples Q3 Q3 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 1.0 1.5 2.0 2.5 3.0 xc_min [mm] 1.0 1.5 2.0 2.5 3.0 xc_min [mm] Q3 Elementary 0.9 - Fitting 0.8 works for 0.7 0.6 both 0.5 samples 0.4 0.3 0.2 0.1 0 1.0 1.5 2.0 2.5 3.0 xc_min [mm]© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 26
  27. 27. Size Descriptors Equivalent Feret diam. 1 spherical diam ┴ to longest Feret diam. 2 Longest diam.© 2010 HORIBA, Ltd. All rights reserved.
  28. 28. Shape: Aspect Ratio Feret diam. 1Aspect ratio Fer= shortest diam et d longest diam ┴ to longest iam .2= ┴ to longest diam longest diam Longest diam.= shortest Feret diam longest Feret diam= three different numbers! © 2010 HORIBA, Ltd. All rights reserved.
  29. 29. More Shape Descriptors© 2010 HORIBA, Ltd. All rights reserved.
  30. 30. Specification with Measurement Error Must tighten internal spec by lab error % Then product always within performance specification Allowance for Allowance for measurement measurement uncertainty uncertainty http://www.spcpress.com/pdf/Manufacturing_Specification.pdf, By David Wheeler© 2010 HORIBA, Ltd. All rights reserved.
  31. 31. Effect of Number of Particles Counted 20,000 particles 200 particles “holes” in distribution second population missed But d10, d50 &d90 may appear similar© 2010 HORIBA, Ltd. All rights reserved.
  32. 32. Two Kinds of Standard Deviation! Sample standard deviation is a property of the sample. It is the width of the size distribution. Measurement standard deviation is the deviation between results from different measurements. It is a result of the measurement and is affected by the sample standard deviation.© 2010 HORIBA, Ltd. All rights reserved.
  33. 33. More particles for more accuracy. dv10 dv50 dv90 90 80 70 5000 60 50 40 30 Assume 49.833 is “correct” 20 dv50, 10 0.95 x 49.833 = 47.34 is 0 within 95%, 47.463 achieved by 5000 00 87 0 0 0 0 0 0 0 00 00 00 10 20 30 50 00 00 00 00 71 10 20 50 counts! 10 20 50 10 17 Use this to control precision of your data (and not spend extra time on precision you don’t need.© 2010 HORIBA, Ltd. All rights reserved.
  34. 34. More particles means more accuracy. 5000 And, it’s really easy to flow a lot of particles through a dynamic image analyzer!© 2010 HORIBA, Ltd. All rights reserved.
  35. 35. Divide Large Data Set into Smaller Sets Error bars are one standard deviation from repeated measurements of the same number of particles from different parts of the sample. The error bars get smaller as you evaluate more particles.© 2010 HORIBA, Ltd. All rights reserved.
  36. 36. CAMSIZER Advantages Measurement Results Measurement of very broad particle distributions (due to speed and two cameras) Direct particle definition by width (analogue to sieving) by length or projection surface Two camera system for more accuracy and reproducibility Easy operation Fail-safe, robust Ideal for particle shape analyses Measurement of density, counting of particles© 2010 HORIBA, Ltd. All rights reserved. © Retsch Technology GmbH 36
  37. 37. Static or Dynamic Image Analysis? Dynamic Broad size distributions (since it is easier to obtain data from a lot of particles) Samples that flow easily (since they must be dropped in front of camera) Powders, pellets, granules Static Samples that are more difficult to disperse (there are more methods for dispersing the samples) Samples that are more delicate Pastes, sticky particles, suspensions© 2010 HORIBA, Ltd. All rights reserved.
  38. 38. Watch out for: Sample preparation Image quality Measure enough particles© 2010 HORIBA, Ltd. All rights reserved.
  39. 39. Dynamic Image Analysis is good for: Replacing Sieves Size Shape Supplementing other techniques© 2010 HORIBA, Ltd. All rights reserved.
  40. 40. Questions? www.horiba.com/us/particle Jeffrey Bodycomb, Ph.D. Jeff.Bodycomb@Horiba.com 732-648-3431© 2010 HORIBA, Ltd. All rights reserved.

×