11
value ranges. It helps maintain data consistency across dif-
ferent images. In addition, histogram equalization was used
to improve contrast and enhance the details in an image by
redistributing the pixel intensity value in an image to cover
a broader and more balanced range. It is particularly useful
for enhancing the texture feature in images intended for
GLCM analysis. Normalization and histogram equalization
were combined to make texture patterns more distinguish-
able and consistent across different images, helping ensure
that texture features are calculated accurately and reliably.
The features for the preprocessed coal images are also
presented in Figure 9. It can be found by comparing the
features with and without preprocessing that, except the
negligible influence on energy value, the preprocessing step
can significantly reduce the variation of the features. This
makes the calculated features more consistent regardless of
the camera settings. However, the comparison of the fea-
tures between BC and BBC shows that the preprocessing
step makes each calculated feature, such as deviation and
mean, show similar distribution with close median values.
This makes the features difficult to classify. The features still
show differences after preprocessing for correlation and
contrast (dissimilarity). Although the preprocessing step
makes the calculated features closer, there are still differ-
ences for correlation and contrast (dissimilarity), making
them still the most important features for the classification
of BC and BBC.
SUMMARY AND CONCLUSIONS
The results from this study show that the coal images of dif-
ferent lithotypes have different textures and the difference
in image texture can be quantified with GLCM method.
Although the GLCM features can be affected by the reso-
lution and camera settings, they can be used effectively to
classify the coal lithotypes. It demonstrates the potential of
coal lithotype classification with rib photos, reducing the
impact of human influence and easing the data collection
process for rib stability analysis.
GLCM method was used to analyze the texture of BC
and BBC images. Based on the recommended minimum
resolution of 1 pixel/mm in a previous study, the GLCM
parameters, namely direction (angle), distance and patch
size, were optimized based on the BC and BBC images.
An angle value of 0 degree, a distance value of 1, and a
patch size of 50×50 mm were used first for the analyses. The
box plots of the GLCM features show that the texture of
BC has higher contrast and dissimilarity levels on average,
indicating that the image texture of BC has significantly
local variation in pixel intensities, resulting in sharp transi-
tions. At the same time, the textures of BC show gener-
ally lower correlation, energy, and homogeneity levels than
those of BBC, indicating that BC images have more ran-
dom, diverse, and heterogeneous textures. In addition, the
GLCM features for BC, in general, have a smaller spread
Figure 9. Box plots of various GLCM features with and without preprocessing
value ranges. It helps maintain data consistency across dif-
ferent images. In addition, histogram equalization was used
to improve contrast and enhance the details in an image by
redistributing the pixel intensity value in an image to cover
a broader and more balanced range. It is particularly useful
for enhancing the texture feature in images intended for
GLCM analysis. Normalization and histogram equalization
were combined to make texture patterns more distinguish-
able and consistent across different images, helping ensure
that texture features are calculated accurately and reliably.
The features for the preprocessed coal images are also
presented in Figure 9. It can be found by comparing the
features with and without preprocessing that, except the
negligible influence on energy value, the preprocessing step
can significantly reduce the variation of the features. This
makes the calculated features more consistent regardless of
the camera settings. However, the comparison of the fea-
tures between BC and BBC shows that the preprocessing
step makes each calculated feature, such as deviation and
mean, show similar distribution with close median values.
This makes the features difficult to classify. The features still
show differences after preprocessing for correlation and
contrast (dissimilarity). Although the preprocessing step
makes the calculated features closer, there are still differ-
ences for correlation and contrast (dissimilarity), making
them still the most important features for the classification
of BC and BBC.
SUMMARY AND CONCLUSIONS
The results from this study show that the coal images of dif-
ferent lithotypes have different textures and the difference
in image texture can be quantified with GLCM method.
Although the GLCM features can be affected by the reso-
lution and camera settings, they can be used effectively to
classify the coal lithotypes. It demonstrates the potential of
coal lithotype classification with rib photos, reducing the
impact of human influence and easing the data collection
process for rib stability analysis.
GLCM method was used to analyze the texture of BC
and BBC images. Based on the recommended minimum
resolution of 1 pixel/mm in a previous study, the GLCM
parameters, namely direction (angle), distance and patch
size, were optimized based on the BC and BBC images.
An angle value of 0 degree, a distance value of 1, and a
patch size of 50×50 mm were used first for the analyses. The
box plots of the GLCM features show that the texture of
BC has higher contrast and dissimilarity levels on average,
indicating that the image texture of BC has significantly
local variation in pixel intensities, resulting in sharp transi-
tions. At the same time, the textures of BC show gener-
ally lower correlation, energy, and homogeneity levels than
those of BBC, indicating that BC images have more ran-
dom, diverse, and heterogeneous textures. In addition, the
GLCM features for BC, in general, have a smaller spread
Figure 9. Box plots of various GLCM features with and without preprocessing