In this paper

we propose a novel semi-automatic ribbon road extraction scheme which combines angular texture signature and profile matching.The angular texture signature was proposed by Haverkamp in 2002 for extracting straight road structure from IKONOS satellite imagery.In this method

a rectangular template region rotates around a specific point with certain angle intervals

and computes texture signature of all points in this template region at each position of rotation.The basic texture signature is variance and the entropy also to be tested in this paper.A profile means that in the centerline points

perpendicular to the direction of the road

we paint a straight line segment

and extract gray values of various points along the line segment.And the straight line segment should be longer than the road width. Profile matching calculates the least square values between profile line model and the profile in the predicted points.To a certain extent

we offset the profile in the predicted point of and calculate values of at the least square each offset position

we believe the corresponding offset with the minimum value should be possible point of road centerline

therefore

the final road centerline can be determined through predicted points and the offset least square values.In the scheme

user inputs the initial position

direction

and width of the road firstly.The initial road centerline points are predicted based on angular texture signature

and road center trajectory model is created using parameters of parabola.The multiple linear regression method is used for acquiring the road trajectory parameters

and we utilize these parameters and the initial centerline points coordinates to compute the initial mean of the curvature.Then we make use of the method of curvature change to verify the points in trajectory.By comparing the value and the given threshold

we’re got following conclusion: if it exceeds the limit

the verification fails

a profile matching algorithm will be used to re-predict and decide the position of center points again;if the value doesn’t exceed the limit

the verification succeeds

we believe the centerline points predicted in angular texture signature is the final position.In the algorithm

the manual intervention is necessary especially for complex situation.After the verification

we can achieve the final centerline points and add them to the trajectory.If these two algorithms both fail

we could decide the final centerline point completely manually.In this paper

we implement the above mentioned scheme and build a prototype system using Visual C++

and verify the extraction result of ribbon roads with appropriate width in QuickBird and IKONOS imagery.In the experiment 1

the road has uniform texture

and the contrast to background is obvious

but no obvious obstructions on the road.In this situation

the scheme extracts the whole road automatically without any manual intervention.In the experiment 2

the ribbon road extraction from IKONOS is generally successful and need a few manual intervention.In the experiment 3

a rather complex situation with several roads in an digital aerial image

we extract the main roads automatically and some junctions of the roads needs semi-automatic extraction

but the final result is satisfactory.We also utilize experiments to compare our scheme with traditional profile matching semi-automatic extraction algorithm and Snakes algorithm.After these experimental verification

it proves this semi-automatic ribbon road extraction scheme can achieve good results.