Customer Solutions
U.S. Army Objectively Measures Respiratory Protective Mask Lens Fogging with IMAQ and LabVIEW
Author(s):
Karen M. Coyne, U.S. Army Edgewood
Industry:
Government/Defense
Product:
LabVIEW, Vision
The Challenge:
Developing an objective method to assess respiratory protective mask lens fogging.
The Solution:
Creating a LabVIEW program using IMAQ Vision Builder that relates Snellen visual acuity to changes in an acquired image due to lens fogging.
Introduction
The current method for quantifying respiratory protective mask lens fogging relies on subjective observations of fogging during controlled environmental tests and field reports of lens fogging during uncontrolled operational testing. These procedures have several drawbacks - the most significant is that the data rely heavily upon the subjective opinions of both test personnel and test participants. This subjective data cannot gauge how lens fogging impacts visual parameters like visual acuity and have only limited use in evaluating a mask lens design feature, such as an antifog coating. An objective measure of fogging of mask eye lenses would overcome these problems.
One prospective concept was to relate characteristics of a captured video image to Snellen visual acuity. The idea was that the pixel gray levels in the video image would change as the lenses fogged. We hoped to be able to relate those changes in the gray levels to Snellen visual acuity.
Developing the Image Processing System
We placed a miniature, black and white video camera in the left eye-socket of a metal headform. Using a Pentium computer with IMAQ Vision Builder and LabVIEW, we acquired and processed the image captured from a PCI-1407 board. We modified acuity using safety goggles with Bangerter occlusion foils of varying Snellen acuities. The foils attached to the lenses with water and were available with acuities of 20/20, 20/25, 20/30, 20/50, 20/70, 20/100, 20/200, 20/300, and occluded. We placed a target in front of the headform. We tested several targets including large, medium, and small circles of varying gray intensities; groups of lines of varying thicknesses and spacing; and several video test patterns. We adjusted two photography lights to obtain minimal shadowing in the video image. We placed a pair of goggles on the headform and obtained an image. Then, we obtained three images for each of the visual acuities and each of the targets.
Processing the Image
Using National Instruments IMAQ Vision Builder, we analyzed the captured images. We applied various processing techniques, such as image and line histograms, filtering, thresholding, edge detection, and circle detection to the images alone and in different combinations. With IMAQ Vision Builder, we quickly tested different targets and processing techniques.
As a result of this initial testing, the most promising target consisted of one large, one medium, and five small circles of varying gray colors. Then, we related the presence or absence of a circle in the captured video image to Snellen visual acuity. The idea was that we would not be able to detect small light gray circles as acuity worsened while we could detect a large black circle for all but the occluded goggles.
We obtained five images of the selected target for each visual acuity. IMAQ Vision Builder evaluated processing techniques. The developed method involved a sequence of seven image-processing techniques including region of interest (ROI), filter, threshold, remove border objects, remove small objects, fill holes, and particle detection. The ROI selected the circle of interest. We applied a filter to the image to decrease the shadowing and sharpen the image. Then, we applied the threshold to convert the image into binary objects, so we could apply the Basic Morphology processes, such as particle analysis. We removed border objects and small objects that appeared due to shadowing, and we filled holes in the image. Finally, we used particle detection to count the number of circles in the image.
Using the IMAQ Vision Builder software, we converted the processing sequence, or script to a LabVIEW program. This aspect of Vision Builder was particularly useful in decreasing programming time. We modified and used the LabVIEW program to process future images. While we used the same sequence for each visual acuity, the ROI and threshold were different. Changing the threshold range allowed us to use some of the circles to check two acuities. The ROI and threshold range were set in the program. The LabVIEW program determined which circles were present and which were absent from the image. The program used a process of elimination to determine the visual acuity.
The ROI and threshold were set, and we checked the first circle. If the program did not detect circle, vision occluded and the processing stopped. If the program detected the circle, the vision was 20/300 or better. The ROI and threshold were changed and the program resumed processing the next circle. If the program did not detect the circle, the acuity was 20/300 and the processing stopped. If the program detected the circle, the acuity was 20/200 or better. The ROI and threshold were set and the next circle examined. This procedure continued through the 20/20 acuity. A simple user interface displayed the Snellen visual acuity and the acquired image. Preliminary testing showed accurate, repeatable results.
Further Development
Future development of this test method will rely heavily on the IMAQ Vision Builder software. We plan to install a second camera in the head form and analyze a sequence of images. These changes will allow us to assess fogging of an entire lens over time. We will test the final technique using a head form that simulates the temperature and sweat rate of a human head. Once we have accomplished these goals, we expect to have a system that can objectively assess the fogging response of respiratory protective masks and protective goggles over time. We will determine the location and severity of fogging, expressed as a Snellen acuity.
Conclusion
LabVIEW and IMAQ Vision Builder allowed us to quickly acquire and process images obtained through "fogged" eye lenses. These software packages helped us show that we could relate changes in pixel characteristics, resulting from lens fogging, to Snellen visual acuity. Without these programs, we would not have known of the numerous options available for image analysis, and development time of our application would have proceeded at a significantly slower pace.
For more information, contact:
Karen M. Coyne, Ph.D.
U.S. Army Edgewood Chemical Biological Center
Tel: 410-436-6520
Fax: 410-436-3141
E-mail: Karen.coyne@sbccom.apgea.army.mil
karen_coyne.pdf
View the entire user solution in Adobe Acrobat PDF format.
