Algorithm 1 Road_Extraction() algorithm.

Require Segmented image, existing road centre-line, starting point

Ensure Road segments

1:

Make the agent traverse through point P(i, j)

2:

Extract segment containing of point P(i, j) in segmented image

3:

Perform the thinning process in extracted segment

4:

Find end points E(i, j) of the VGI line segment corresponding to the traversed segment

5:

for each end points do

6:

 Consider the neighbouring pixels E(i + k, j +1) of E(i, j), where k, l = ±1,2

7:

 Among these neighbouring pixels, find only those pixels E(i + k, j + l) such that these pixels belong to the different segment other than the current traversing segment

8:

 Validate the neighbouring segment as road either based on mean color deviation, where threshold is 1.5 times of previous segment or, by considering it as shadow (NDVI < 0) or trees(NDVI > 0.3)

9:

  if neighbouring segment is validated as existing road: then

10:

 Consider neighbouring pixel E(i ± k, j ± l) as road if it is in direction of known existing road

11:

 Extract segment of neighbouring pixel E(i + k, j + l)

12:

 REPEAT the entire process for the obtained segment

13:

Combine all extracted road segment