Row guidance parameterization with hough transform
Systems and techniques for row guidance parameterization with Hough transform are described herein. An electronic representation of a field (ERF) can be received. The ERF can include a set of feature sets including one of a set of crop row features or a set of furrow features. A first parameter space can be produced by applying a slope-intercept Hough transform (SLIHT) to members of a feature set. Peaks in the first parameter space can be identified. A second parameter space can be produced by application of the SLIHT to the peaks. A vanishing point can be calculated based on a vanishing point peak in the second parameter space. A track-angle error can be calculated from the vanishing point.
1. A non-transitory machine-readable medium including instructions that, when executed by a machine, cause the machine to perform operations comprising:
receiving an image with a set of features;
detecting a plurality of lines corresponding to the set of features;
calculating a vanishing point from the plurality of lines;
projecting a projected line from a point on the image representing a vehicle to the vanishing point;
calculating an angle error between the projected line and a vehicle direction line from the point on the image, wherein the vehicle direction line corresponds to a heading of the vehicle; and
providing the angle error to a steering module of a vehicle to which the image capture device is mounted.
2. The machine-readable medium of claim 1 , wherein detecting the plurality of lines includes using a line voting mechanism.
3. The machine-readable medium of claim 2 , wherein the voting mechanism includes:
identifying a candidate line from a predetermined pool of candidate lines;
applying the candidate line to the set of features; and
accumulating votes for the candidate line for features in the set of features that correspond to the candidate line.
4. The machine-readable medium of claim 3 , wherein candidate pool of candidate lines are defined by a parameter space that describe the candidate lines.
5. The machine-readable medium of claim 1 , wherein calculating the vanishing point from the plurality of lines includes:
mapping a first line and a second line from the image to respective features in a second feature set;
isolating a third line in the second feature set upon which both respective features lie;
mapping the third line to a point in a third feature set; and
translating the point in the third feature set to the vanishing point in the image.
6. The machine-readable medium of claim 1 , comprising calculating a cross-track distance from the machine and a line parallel to one of the plurality of lines and providing the cross-track distance to the steering module.
7. A method comprising:
receiving, by a scene module, an image with a set of features;
detecting, by a transform module, a plurality of lines corresponding to the set of features;
calculating a vanishing point from the plurality of lines;
creating a projected line from a point on from a point on the image representing a vehicle to the vanishing point
calculating, by a track-angle error module an angle error between the projected line and a vehicle direction line from the point on the image, wherein the vehicle direction line corresponds to a heading of the vehicle; and
providing the angle error to a steering module of a vehicle to which the image capture device is mounted.
8. The method of claim 7 , wherein detecting the plurality of lines includes using a line voting mechanism.
9. The method of claim 8 , wherein the voting mechanism includes:
identifying a candidate line from a predetermined pool of candidate lines;
applying the candidate line to the set of features; and
accumulating votes for the candidate line for features in the set of features that correspond to the candidate line.
10. The method of claim 9 , wherein candidate pool of candidate lines are defined by a parameter space that describe the candidate lines.
11. The method of claim 7 , wherein calculating the vanishing point from the plurality of lines includes:
mapping afirst line and a second line from the image to respective features in a second feature set;
isolating a third line in the second feature set upon which both respective features lie;
mapping the third line to a point in a third feature set; and
translating the point in the third feature set to the vanishing point in the image.
12. The method of claim 7 , comprising calculating a cross-track distance from the machine and a line parallel to one of the plurality of lines and providing the cross-track distance to the steering module.
13. A system comprising:
a scene module to receive an image with a set of features;
a transform module to:
detect a plurality of lines corresponding to the set of features; and
calculate a vanishing point from the plurality of lines; and
a track-angle error module to:
create a projected line from a point on the image representing a vehicle to the vanishing point
calculate an angle error between the projected line and a vehicle direction line from the point on the image, wherein the vehicle direction line corresponds to a heading of the vehicle; and
provide the angle error to a steering module of a vehicle to which the image capture device is mounted.
14. The system of claim 13 , wherein to detect the plurality of lines includes the transform module arranged to use a line voting mechanism.
15. The system of claim 14 , wherein the voting mechanism includes the transform module arranged to:
identify a candidate line from a predetermined pool of candidate lines;
apply the candidate line to the set of features; and
accumulate votes for the candidate line for features in the set of features that correspond to the candidate line.
16. The system of claim 15 , wherein candidate pool of candidate lines are defined by a parameter space that describe the candidate lines.
17. The system of claim 13 , wherein to calculate the intersection of the first line and the second line includes the transform module arranged to:
map a first line and a second line from the image to respective features in a second feature set;
isolate a third line in the second feature set upon which both respective features lie;
map the third line to a point in a third feature set; and
translate the point in the third feature set to the vanishing point in the first image.
18. The system of claim 13 , comprising a cross-track distance module to calculate a cross-track distance from the machine and a line parallel to one of the plurality of lines.
19. The system of claim 18 , wherein the cross-track distance module is to provide the cross-track distance to the steering module.