Linear correspondence assessment
First Claim
Patent Images
1. A feature extraction and evaluation system, comprising:
- an extraction system configured to produce extracted features from input images; and
an evaluation system configured to find a correspondence between reference features and the extracted features.
2 Assignments
0 Petitions
Accused Products
Abstract
A system and method for feature extraction and evaluation include an extraction system configured to produce extracted features from input images, and an evaluation system configured to find a correspondence between reference features and the extracted features.
32 Citations
34 Claims
-
1. A feature extraction and evaluation system, comprising:
-
an extraction system configured to produce extracted features from input images; and
an evaluation system configured to find a correspondence between reference features and the extracted features. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12)
-
-
13. A method for extracting and evaluating features, comprising:
-
extracting features from input images to obtain extracted features; and
finding a correspondence between predefined reference data and the extracted features. - View Dependent Claims (14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24)
-
-
25. A method of rasterizing and buffering, comprising:
-
assigning a first value to all pixels in an existence layer in which an extracted vector line is present, the first value being different from a second value assigned to all pixels in which no extracted vector line is present;
assigning a length of the extracted vector line to all pixels in a length layer that correspond to pixels in the existence layer having the first value;
generating a buffer area around a reference vector line associated with the extracted vector line; and
computing a distance from a center of each pixel in the buffer area to a nearest point on the reference vector line and storing the distance in a respective pixel of a distance layer; and
computing an orientation of the reference vector line for each pixel in the buffer area and storing the orientation in a respective pixel of an orientation layer. - View Dependent Claims (26, 27)
-
-
28. A method of assessing linear correspondence, comprising:
-
setting an image frame so that the image frame is configured by a map extent of an overlapping area of an input evaluation dataset and an input reference dataset, by a cell size of the image, and by offsets of the evaluation dataset;
generating an existence raster layer, a length raster layer, a distance raster layer and an orientation raster layer from each dataset using rasterizing by assigning input vector lines in an existence layer with a second value for negative pixels and a first value for positive pixels and recording a length of the vector lines in a length layer, and using buffering by generating a buffer area around a line segment and computing a distance and an orientation at pixels within the buffer area and assigning the distance and orientation to the distance and orientation layers; and
performing matching in two directions, one direction being from reference to evaluation and the other direction being from evaluation to reference, wherein the matching in the direction reference to evaluation uses the existence layer and the length layer from the evaluation dataset and the matching in the direction evaluation to reference uses the distance layer and the orientation layer. - View Dependent Claims (29, 30, 31, 32, 33, 34)
-
Specification