Method and system for rectifying images

1. A method for rectifying an image, comprising:

obtaining, by a computing device, an original stereo image pair;

parallelizing, by the computing device, epipolar lines corresponding to the original stereo image pair to obtain a first transformed stereo image pair; and

collinearizing, by the computing device, epipolar lines corresponding to the first transformed stereo image pair to obtain a second transformed stereo image pair,

wherein obtaining of the original stereo image pair further comprises creating a stereo image point pair set by the computing device comprises:

obtaining two angular point sets through angular point extraction;

establishing mapping between angular points in the two angular point sets through feature matching, and

creating the stereo image point pair set comprising two corresponding angular points.

2. The method for rectifying images according to claim 1 , wherein the parallelizing of the epipolar lines corresponding to the original stereo image pair comprises:

determining, by the computing device, a basic plane according to a baseline of a stereo vision system and a plane where the original stereo image pair is located;

projecting, by the computing device, the original stereo image pair onto the basic plane to obtain a projected stereo image pair; and

parallelizing, by the computing device, epipolar lines corresponding to the projected stereo image pair.

3. The method for rectifying images according to claim 1 , wherein the parallelizing of the epipolar lines corresponding to the original stereo image pair comprises:

randomly selecting a stereo image point pair from the stereo image point pair set; calculating a basic matrix according to the stereo image point pair; and

parallelizing the epipolar lines corresponding to the original stereo image pair according to the basic matrix.

4. The method for rectifying images according to claim 1 , further comprising:

centralizing the second transformed stereo image pair according to centers of the original stereo image pair to obtain a third transformed stereo image pair.

5. The method for rectifying images according to claim 4 , further comprising:

calculating values of coordinates of four vertexes in the second transformed stereo image pair on an X-axis respectively, with maximum and minimum values removed;

calculating values of coordinates of four vertexes in the second transformed stereo image pair on a Y-axis respectively, with maximum and minimum values removed; and

calculating maximum and minimum values of coordinates of vertexes in a trimmed area from the second transformed stereo image pair to obtain a fourth transformed stereo image pair.

6. A system for rectifying images, comprising a non-transitory storage medium accessible to a processor, the non-transitory storage medium comprising:

an image pair obtaining module, configured to obtain, by the system, an original stereo image pair;

an epipolar line parallelizing module, connected to the image pair obtaining module and configured to, by the system, parallelize epipolar lines corresponding to the original stereo image pair to obtain a first transformed stereo image pair;

an epipolar line collinearizing module, connected to the epipolar line parallelizing module and configured to, by the system, collinearize epipolar lines corresponding to the first transformed stereo image pair to obtain a second transformed stereo image pair; and

an image point pair set creating module, connected to the image pair obtaining module and the epipolar line parallelizing module respectively and adapted to create and save a stereo image point pair set,

wherein the image point pair set creating module is further configured to obtain two angular point sets through angular point extraction;

establish mapping between angular points in the two angular point sets through feature matching, and

create the stereo image point pair set comprising two corresponding angular points.

7. The system for rectifying images according to claim 6 , wherein the epipolar line parallelizing module comprises:

a basic plane determining module, adapted to determine a basic plane according to a baseline of a stereo vision system and a plane where the original stereo image pair is located;

a projecting module, connected to the basic plane determining module and adapted to project the original stereo image pair onto the basic plane to obtain a projected stereo image pair; and

a first parallelizing module, connected to the projecting module and adapted to parallelize epipolar lines corresponding to the projected stereo image pair.

8. The system for rectifying images according to claim 1 , wherein the epipolar line parallelizing module comprises:

a basic matrix determining module, adapted to randomly select a stereo image point pair from the stereo image point pair set, and calculate a basic matrix according to the stereo image point pair; and

a second parallelizing module connected to the basic matrix determining module and adapted to parallelize the epipolar lines corresponding to the original stereo image pair according to the basic matrix.

9. The system for rectifying images according to claim 6 , further comprising:

an image centralizing module, connected to the epipolar line collinearizing module and adapted to centralize the stereo image pair transformed for the second time according to centers of the original stereo image pair to obtain a third transformed stereo image pair.

10. The system for rectifying images according to claim 9 , further comprising: an image trimming module, connected to the epipolar line collinearizing module and adapted to respectively calculate values of coordinates of four vertexes in the second transformed stereo image pair on an X-axis, with maximum and minimum values removed; respectively calculate values of coordinates of four vertexes in the second transformed stereo image pair on a Y-axis, with maximum and minimum values removed; and obtain maximum and minimum values of coordinates of vertexes in a trimmed area from the second transformed stereo image pair to obtain a fourth transformed stereo image pair.

11. The system for rectifying images according to claim 9 , further comprising: an image trimming module, connected to the image centralizing module and adapted to respectively calculate values of coordinates of four vertexes in the second transformed stereo image pair on an X-axis, with maximum and minimum values removed; respectively calculate values of coordinates of four vertexes in the second transformed stereo image pair transformed on a Y-axis, with maximum and minimum values removed; and obtain maximum and minimum values of coordinates of vertexes in a trimmed area from the second transformed stereo image pair to obtain a fourth transformed stereo image pair.

12. A computing device comprising a processor and a non-transitory storage accessible to the processor, configured to:

obtain an original stereo image pair;

parallelize epipolar lines corresponding to the original stereo image pair to obtain a first transformed stereo image pair; and

collinearize epipolar lines corresponding to the first transformed stereo image pair to obtain a second transformed stereo image pair,

wherein the computing device obtains the original stereo image pair by creating a stereo image point pair set through angular point extraction and feature matching,

wherein the computing device, further configured to:

obtain two angular point sets through angular point extraction;

establish mapping between angular points in the two angular point sets through feature matching, and

create the stereo image point pair set comprising two corresponding angular points.