Table 1.

Summary of descriptors.

Shape
Area	A=nPixels
Perimeter	$P=\sqrt{{\left(x_i-x_{i-1}\right)}^2+{\left(y_i-y_{i-1}\right)}^2}$
Roundness	R=4π(A/P2)
Compactness	C=A/P2
First-order texture
Average	μ=1/ij∑i,jp(i, j)
Median	m=L+I((N/2) − F/f)
Variance	σ 2=1/ij∑i,j((p(i, j) − μ))
Standard deviation	$\sigma =\sqrt{1/ij{\sum }_{i,j}\left(p\left(i,j\right)-\mu \right)}$
Entropy	S=−∑i,jp(i, j)log p(i, j)
Second-order texture
Contrast descriptor	CM=∑i,j|i − j|2c(i, j)
Correlation	r=∑i,j(i − μci)(j − μcj)c(i, j)/σciσcj
Energy	e=∑i,jc(i, j)2
Local homogeneity	HL=∑i,jc(i, j)/(1+|i − j|)
Relation context
Euclidian distance	$\text{ED}=\sqrt{{\left(V_r-V_l\right)}^2}$
Bhattacharyya distance	$\text{BD}=\sqrt{\left(V_r\times V_l\right)}$
Difference	D=abs(Vr − Vl)