Image Capturing and Analysis
Raw Digital Images (Gray Scale)—Images of the sealant samples were taken under an optimized diffuse illumination at a constant illuminance. This type of illumination enhances the chromatic differences of the sealant surfaces. Two sets of images were captured per each specimen: the weathered surface for deterioration evaluation and the unweathered surface on the lateral sample side for control (reference) image. While the use of an original unexposed specimen as the control image is desirable, this method was chosen, as no other reference sample was available. Figure 3 shows the raw digital images of the reference (unweathered) and weathered surfaces of the sealant samples.
Overall Surface Texture Change—This property is a quick, yet robust, means of measuring the overall surface deterioration. It is measured based on tonal variations caused by surface defects in digital images. For example, surface cracks and their interaction with incident light cause severe tonal variation in the captured image.
Cracking (Crazing) Analysis—Generally, the severity of cracking or crazing can be characterized in several ways, e. g., by determining density of cracks (amount of cracks shown in a given surface), nor-
FIG. A—Digital images of weathered sealant samples and overall surface texture change index.
malized crack density (based on skeletonized crack structure), crack thickness (width of crack), crack orientation (direction of cracks), etc. In sealants, density of cracks and crack thickness combined appropriately describe cracking severity in a quantitative manner.
Cracks, or other defects that are defined by boundaries, require that the images be converted to a black and white format. In the black and white image, cracks are shown as black pixels and the background as white pixels. In describing crack width, the term “main crack width” is used. This parameter is calculated as average crack width plus standard deviation. Figure 5 shows the binary (black and white) image of the weathered sealant samples and the calculated crack density and main crack width.
Visual Color (Tonal) Change—The visual color change is measured with respect to the average gray level of an image. With constant illumination condition, the gray level of the specimen surface depicted in a digital image corresponds to the specimen’s visual color on the gray scale. In order to assess the changes from the original condition, the lateral side unweathered surface was used as the gray scale reference. For the specimens with a cracked surface, cracks were subtracted from the image, because the cracks and their interaction with light make the entire specimen surface appear darker than the actual surface color.
Average gray level values of control specimens and exposed specimens were used to calculate the change in visual color—see Fig. 6 for inverted gray images of weathered and unweathered sealant surfaces. Figure 7 illustrates the actual changes in gray levels, while Fig. 8 illustrates the color change, which is calculated based on gray levels of control surfaces minus the gray level of exposed surfaces.
Uniformity of Deterioration and Virtual Human Visual Evaluation—Digital images of a specimen surface often contain more data points than the human eye perceives. Human perception often averages the
FIG. 5—Binary image, crack density and main crack width for weathered sealant samples.
details of a specimen surface and provides an overall judgment of the surface condition. This overall judgment is mostly based on the uniformity of surface deterioration based on their tonal variations. For example, cracked surfaces are by far the most severe damage in materials and their visual tonal variations result in highly nonuniform surfaces. With proper processing of digital images, surface images can be simplified in the same manner as is done by human perception. With the simplified (processed) images, one can digitally mimic human judgment of a surface condition and provide quantitative data that correlate
FIG. 6—Inverted gray images of weathered and unweathered sealant surfaces used in the calculation of color changes.
FIG. 7—Average gray levels of unexposed (control) and exposed sealant surfaces.
with human visual judgment (see Fig. 9 for enhanced and processed images of weathered sealant surfaces).
The images in the left column of Fig. 9 represent “enhanced specimen images.” This enhancement amplifies surface damage and thus the tonal variation is superior to the original images. The images in the right column of Fig. 9 are further processed in order to mimic human perception of surface condition. In this process, similar gray level neighboring pixels are combined to represent simpler images. From these images, the uniformity of surface deterioration is measured based on tonal variation.