Method for automatic alignment of raster data with vector data in a geographic information system
AssigneeGoogle Inc.
InventorXiaqing Wu
Life of this patent
Estimating the term…
Abstract
The present invention relates to methods for aligning raster and vector data. In an embodiment, a raster/vector aligner receives raster data and an approximate vector of a feature within the raster data. The raster/vector aligner generates an edge signal by edge filtering the raster data along a direction of the approximate vector and a smoothness signal by smoothness filtering the raster data along a direction of the approximate vector. The raster/vector aligner combines the edge signal and the smoothness signal into a combined signal which is used to generate a translation vector or a signal weight for the feature within the raster data.
Claims16
A method of aligning raster data and vector data, comprising:
- receiving, by one or more processors, raster data and a set of approximate vectors of vector data associated with features within the raster data;
- identifying, by the one or more processors, directions of the approximate vectors and grouping the approximate vectors into one or more directional chunks based on their directions;
- defining, by the one or more processors, search areas of the raster data for each directional chunk of the one or more directional chunks, each search area of the search areas encompassing at least one approximate vector of the set of approximate vectors;
- detecting, by the one or more processors, a road feature from said each search area and at least one approximate vector;
- generating, by the one or more processors, a translation vector and a confidence ratio from the detected road feature, wherein generating the translation vector comprises calculating the translation vector in accordance with:
- cost
- (
- r
- ,
- α
- )
- =
- ∑
- i
- =
- 1
- I
-
- ∑
- j
- =
- 1
- J
- i
-
- W
- i
- ,
- j
-
- (
- L
- i
- ,
- j
- -
- r
-
-
- cos
-
- (
- α
- -
- θ
- i
- )
- )
- 2
- wherein (r,α) represents the translation vector of length equal to r and direction equal to α, cost(r,α) represents a target cost function, Wi,j represents a signal weight of the jth vector of direction equal to θi, Li,j represents a location of the jth vector of direction equal to θi, I represents a total number of different directions identified within said each directional chunk, Ji represents a total number of possible road feature locations identified for direction θi, and i and j represent summation indices;
- and
- aligning by the one or more processors, the raster data for said each search area with the at least one approximate vector using the translation vector.
A non-transitory computer readable storage medium having instructions stored thereon that, when executed by a processor, cause the processor to perform a method, the method comprising:
- receiving raster data and a set of approximate vectors of vector data associated with features within the raster data;
- identifying directions of the approximate vectors and grouping the approximate vectors into one or more directional chunks based on their directions;
- defining search areas of the raster data for each directional chunk of the one or more directional chunks, each search area of the search areas encompassing at least one approximate vector of the set of approximate vectors;
- detecting a road feature from said each search area and at least one approximate vector;
- generating a translation vector and a confidence ratio from the detected road feature, wherein generating the translation vector comprises calculating the translation vector in accordance with:
- cost
- (
- r
- ,
- α
- )
- =
- ∑
- i
- =
- 1
- I
-
- ∑
- j
- =
- 1
- J
- i
-
- W
- i
- ,
- j
-
- (
- L
- i
- ,
- j
- -
- r
-
-
- cos
-
- (
- α
- -
- θ
- i
- )
- )
- 2
- wherein (r,α) represents the translation vector of length equal to r and direction equal to α, cost(r,α) represents a target cost function, Wi,j represents a signal weight of the jth vector of direction equal to θi, Li,j represents a location of the jth vector of direction equal to θi, I represents a total number of different directions identified within said each directional chunk Ji represents a total number of possible road feature locations identified for direction θi, and i and j represent summation indices;
- and
- aligning the raster data for said each search area with the at least one approximate vector using the translation vector.
Dependencies are inferred from the claim text.