Method and apparatus for localization based on images and map data
First Claim
Patent Images
1. A localization method, comprising:
- generating a first image of an object from an input image;
generating a second image to project the object with respect to a plurality of candidate localization information of a device, based on map data comprising a position of the object;
pooling, from the first image, feature values corresponding to vertices in the second image;
determining scores of the plurality of candidate localization information based on the pooled feature values; and
determining a candidate localization information corresponding to a highest score to be localization information of the devicewherein the determining of the scores comprises;
determining rotational degree of freedom (DOF)s based on long-distance vertices affected below a first threshold by translational DOFs, from among vertices included in the second image;
after the determination of the rotational DOFs, determining a left and right translational DOF based on vanishing point-oriented short-distance vertices affected below a second threshold by a forward and backward translational DOF, from among short-distance vertices excluding the long-distance vertices from the second image; and
after the determination of the left and right translational DOF, determining the forward and backward translational DOF based on non-vanishing point-oriented short-distance vertices excluding the vanishing point-oriented short-distance vertices from the short-distance vertices.
1 Assignment
0 Petitions
Accused Products
Abstract
Localization apparatuses and methods are disclosed where a localization apparatus extracts a feature of an object from an input image, generates an image in which the object is projected with respect to localization information of a device based on map data, and evaluates the localization information based on feature values corresponding to vertices included in a projection image.
12 Citations
30 Claims
-
1. A localization method, comprising:
-
generating a first image of an object from an input image; generating a second image to project the object with respect to a plurality of candidate localization information of a device, based on map data comprising a position of the object; pooling, from the first image, feature values corresponding to vertices in the second image; determining scores of the plurality of candidate localization information based on the pooled feature values; and determining a candidate localization information corresponding to a highest score to be localization information of the device wherein the determining of the scores comprises; determining rotational degree of freedom (DOF)s based on long-distance vertices affected below a first threshold by translational DOFs, from among vertices included in the second image; after the determination of the rotational DOFs, determining a left and right translational DOF based on vanishing point-oriented short-distance vertices affected below a second threshold by a forward and backward translational DOF, from among short-distance vertices excluding the long-distance vertices from the second image; and after the determination of the left and right translational DOF, determining the forward and backward translational DOF based on non-vanishing point-oriented short-distance vertices excluding the vanishing point-oriented short-distance vertices from the short-distance vertices. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)
-
-
18. A localization method, comprising:
-
generating a first image of an object from an input image; generating a second image to project the object with respect to candidate localization information of a device, based on map data comprising a position of the object; segmenting the second image into a long-distance region and a short-distance region based on a first threshold associated with a distance; segmenting the short-distance region into a vanishing point-oriented short-distance region and a non-vanishing point-oriented short-distance region based on a second criterion associated with a vanishing point; and determining degree of freedom (DOF) values included in the candidate localization information through matching between the first image and the regions, wherein the determining of the DOF comprises; determining rotational DOFs based on long-distance vertices affected below a first threshold by translational DOFs, from among vertices included in the second image, after the determination of the rotational DOFs, determining a left and right translational DOF based on vanishing point-oriented short-distance vertices affected below a second threshold by a forward and backward translational DOF, from among short-distance vertices excluding the long-distance vertices from the second image, and after the determination of the left and right translational DOF, determining the forward and backward translational DOF based on non-vanishing point-oriented short-distance vertices excluding the vanishing point-oriented short-distance vertices from the short-distance vertices. - View Dependent Claims (19, 20, 21, 22, 23, 24)
-
-
25. A localization apparatus, comprising:
-
a processor configured to; generate a first image of an object from an input image, generate a second image to project the object with respect to a plurality of candidate localization information of a device, based on map data comprising a position of the object, pool, from the first image, feature values corresponding to vertices in the second image, determine scores of the plurality of candidate localization information based on the pooled feature values; and determine a candidate localization information corresponding to a highest score to be localization information of the device wherein the determining of the scores comprises; determining rotational degree of freedom (DOF)s based on long-distance vertices affected below a first threshold by translational DOFs, from among vertices included in the second image; after the determination of the rotational DOFs, determining a left and right translational DOF based on vanishing point-oriented short-distance vertices affected below a second threshold by a forward and backward translational DOF, from among short-distance vertices excluding the long-distance vertices from the second image; and after the determination of the left and right translational DOF, determining the forward and backward translational DOF based on non-vanishing point-oriented short-distance vertices excluding the vanishing point-oriented short-distance vertices from the short-distance vertices.
-
-
26. A localization apparatus, comprising:
-
a processor configured to; generate a first image of an object from an input image, generate a second image to project the object with respect to candidate localization information of a device, based on map data comprising a position of the object, segment the second image into a long-distance region and a short-distance region based on a first threshold associated with a distance, segment the short-distance region into a vanishing point-oriented short-distance region and a non-vanishing point-oriented short-distance region based on a second criterion associated with a vanishing point; and determine degree of freedom (DOF) values included in the candidate localization information through matching between the first image and the regions, wherein the processor is further configured to; determine rotational degree of freedom (DOF) values based on the long-distance region, after the determination of the rotational DOF, determine a left and right translational DOF based on the vanishing point-oriented short-distance region, and after the determination of the left and right translational DOF, determine a forward and backward translational DOF based on the non-vanishing point-oriented short-distance region.
-
-
27. A localization apparatus comprising:
-
a sensor disposed on a device, and being configured to sense one or more of an image and candidate localization information of the device; a processor configured to generate a first image of an object from the image, generate a second image to project the object with respect to the candidate localization information, based on map data comprising a position of the object, segment the second image into a long-distance region and a short-distance region based on a distance, segment the short-distance region into a vanishing point-oriented short-distance region and a non-vanishing point-oriented short-distance region based on a vanishing point, determine a score of the candidate localization information based on pooling, from the first image, feature values corresponding to vertices in the second image, determine localization information of the device based on the score, and a head-up display (HUD) configured to visualize a virtual object on the map data based on the determined localization information, wherein the processor to determine is configured to; determine rotational degree of freedom (DOF) values based on the long-distance region; after the determination of the rotational DOF, determine a left and right translational DOF based on the vanishing point-oriented short-distance region; and after the determination of the left and right translational DOF, determine a forward and backward translational DOF based on the non-vanishing point-oriented short-distance region. - View Dependent Claims (28, 29, 30)
-
Specification