10
are all concentrated, it can be expected from the KDE plot
that there is less overlapped area, making it easier to classify
these two groups of data. As a result, the increase in resolu-
tion can help capture the difference in textures between BC
and BBC.
Influence of Camera Setting on Coal Lithotype
Classification
The GLCM is a texture analysis method used to extract tex-
ture features from an image and the features are calculated
from the co-occurrence matrix, which represents the spa-
tial relationship of pixel intensities in an image. Different
camera settings, such as shutter speed, F-stop, and ISO,
can affect the GLCM features calculated from photos.
These camera settings impact the way the images are cap-
tured, leading to changes in the pixel values, illumination,
and texture characteristics (Slaker and Mohamed, 2016).
The shutter speed determines the duration of light expo-
sure on a camera’s sensor, and longer shutter speeds cap-
ture more light, resulting in brighter images with reduced
noise. Changes in shutter speed can affect pixel intensity
distribution and impact GLCM features like contrast and
homogeneity. The F-stop, also known as aperture, controls
the amount of light entering the camera’s sensor. Higher
F-stop values (smaller aperture) reduce light, increase depth
of field, and make the image sharper. F-stop changes can
influence GLCM features related to image sharpness. The
ISO value represents the sensitivity of the camera’s sensor
to light. Higher ISO values make the sensor more sensi-
tive and result in brighter images but may introduce more
noise. Changes in ISO can affect GLCM parameters related
to image brightness and noise.
When taking photos for the fresh surfaces on the ribs,
different camera settings were used to explore the influence
of camera exposure on GLCM features. As recommended
by Slaker and Mohamed (2016), a shutter speed between
1/60 s and 1/100 s, a F-stop value between 5 and 8, and
an ISO value between 100 and 400 were selected, and dif-
ferent combinations of these settings were used when tak-
ing rib photos. Four examples were selected for each coal
lithotype for the following analysis and are presented in
Figure 8. Correspondingly, the image features calculated
from the original photos without preprocessing are shown
in Figure 9. The first images for BC and BBC are the ones
that have been used in the analyses in previous sections.
Compared to the first images, visual inspections show that
the fourth image have similar luminance as the first ones
while the second and third ones have higher or lower lumi-
nance than the first ones, resulted from different camera set-
tings. Some common characteristics are observed between
BC and BBC in Figure 9. The images with lower luminance
have lower contrast, dissimilarity, deviation and mean val-
ues and higher energy and homogeneity values, while the
images with higher luminance have higher mean value but
in general, the GLCM features vary with the exposure.
Instead of using the original photos captured with dif-
ferent camera settings, image preprocessing steps, namely
normalization and histogram equalization, were taken to
improve the quality of the images before applying texture
analysis. These steps can help enhance the visibility of tex-
ture patterns and ensure that the image data is suitable for
texture analysis. Image normalization was used to stan-
dardize the pixel values to a specific range, such as [0,1]
or [0,255], to ensure that all images have consistent pixel
(1) (2) (3) (4)
(a) BC
(1) (2) (3) (4)
(b) BBC
Figure 8. Coal images captured with different camera settings
Previous Page Next Page