Table 10.

Textural features extracted using GLCM

Feature name	Equation	Definition
Autocorrelation	$\mathrm{acorr}=\sum _i\sum _j\mathrm{ijp}(i,j)$	Linear dependence in GLCM between same index
Cluster Shade	$C_{\mathrm{shade}}=\sum _i\sum _j{(i+j-{\mu }_x-{\mu }_y)}^{3}p(i,j)$	Measure of skewness or non-symmetry
Cluster Prominence	$C_{\mathrm{prom}}=\sum _i\sum _j{(i+j-{\mu }_x-{\mu }_y)}^{4}p(i,j)$	Show peak in GLCM around the mean for non-symmetry
Contrast	$\mathrm{con}=\sum _{i=1}^{N_g}\sum _{j=1}^{N_g}|i-j{|}^{2}p(i,j)$	Local variations to show the texture fineness.
Correlation	$\mathrm{corr}=\frac{\sum _i\sum _j\left(\mathrm{ij}\right)p(i,j)-{\mu }_x{\mu }_y}{{\sigma }_x{\sigma }_y}$	Linear dependence in GLCM between different index
Difference Entropy	$H_{\mathrm{diff}}=-\sum _{i=0}^{N_g-1}{p}_{x-y}\log \left(p_{x-y}\right(i\left)\right)$	Higher weight on higher difference of index entropy value
Dissimilarity	$\mathrm{diss}=\sum _i\sum _j|i-j|p(i,j)$	Higher weights of GLCM probabilities away from the diagonal
Energy	$E=\sum _i\sum _j{p(i,j)}^2$	Returns the sum of squared elements in the GLCM
Entropy	$H=-\sum _i\sum _{j}p(i,j)\log \left(p\right(i,j\left)\right)$	Texture randomness producing a low value for an irregular GLCM
Homogeneity	$\mathrm{homom}=\sum _i\sum _j\frac{1}{1+{(i-j)}^2}p(i,j)$	Closeness of the element distribution in GLCM to its diagonal
Information Measures 1	I M 1=(1−e x p[−2.0(H xy−H)])0.5	Entropy measures
Information Measures 2	$I{M}_2=\frac{\mathrm{Entropy}-H_{\mathrm{xy}2}}{\text{MAX}(H_x,H_y)}$	Entropy measures
Inverse Difference	$\mathrm{IDN}=\sum _i\sum _j\frac{p(i,j)}{1+\frac{|i-j|}{N_g}}$	Inverse Contrast Normalized
Normalized
Inverse Difference Moment	$\mathrm{IDMN}=\sum _i\sum _j\frac{p(i,j)}{1+\frac{{(i-j)}^2}{N_g}}$	Homogeneity Normalized
Normalized
Maximum Probability	$P{r}_{\max }=\underset{(x,y)}{\text{MAX}}p(i,j)$	Maximum value of GLCM
Sum average	${\mu }_{\mathrm{sum}}=\sum _{i=2}^{2N_g}i{p}_{x+y}\left(i\right)$	Higher weights to higher index of marginal GLCM
Sum Entropy	$H_{\mathrm{sum}}=-\sum _{i=2}^{2N_g}{p}_{x+y}\log \left(p_{x+y}\right(i\left)\right)$	Higher weight on higher sum of index entropy value
Sum of Squares: Variance	${\sigma }_{\mathrm{sos}}=\sum _i\sum _j{(i-\mu )}^{2}p(i,j)$	Higher weights that differ from average value of GLCM
Sum of Variance	${\sigma }_{\mathrm{sum}}=\sum _{i=2}^{2N_g}(i-H_{\mathrm{sum}})p_{x+y}\left(i\right)$	Higher weights that differ from entropy value of marginal GLCM

(i,j) represent rows and columns respectively, N _g is number of distinct grey levels in the quantised image, p(i,j) is the element from normalized GLCM matrix p _x(i) and p _y(j) are marginal probabilities of matrix obtained by summing rows and columns of GLCM respectively i.e. $p_x\left(i\right)=\sum _{j=1}^{N_g}p(i,j)$, $p_y\left(j\right)=\sum _{i=1}^{N_g}p(i,j)$, $p_{x+y}\left(k\right)=\sum _{i=1}^{N_g}\sum _{j=1}^{N_g}p(i,j),k=i+j-1=1,2,3,\mathrm{....},2N_g$ and $p_{x-y}\left(k\right)=\sum _{i=1}^{N_g}\sum _{j=1}^{N_g}p(i,j),k=|i-j|+1=1,\mathrm{....},N_g$, H _x and H _y and entropies of p _x and p _y respectively, $H_{\mathrm{xy}}=-\sum _i\sum _j{p}_x\left(i\right)p_y\left(j\right)\log \left(p_x\right(i\left)p_y\right(j\left)\right)$, $H_{\mathrm{xy}2}=-\sum _i\sum _{j}p(i,j)\log \left(p_x\right(i\left)p_y\right(j\left)\right)$