Process for recognition of lane markers using image data
First Claim
Patent Images
1. A process for recognition of vehicle lane markings from image data, comprising:
- evaluating morphological characteristics of vehicle lane markings using a priori knowledge, andusing a matched-filter in order to extract image points, which are associated with vehicle lane markers, by measuring the average gray value of the background in the environment of the position to be examined, and evaluating an image point, which is potentially to be associated with the vehicle lane marking, on the basis of a comparison between background noise, the average gray value in the environment, and a gray value of the position to be examined.
2 Assignments
0 Petitions
Accused Products
Abstract
In the inventive process the morphological characteristics of point-shaped vehicle lane markers are evaluated on the basis of a priori knowledge. For this, those estimation processes are particularly useful, which are based on a Kalman Filter. After the selection of suitable ROI a Matched Filter is employed, in order to better extract from the background the image points associated with a vehicle lane marker. The inventive process is obviously not limited to point-shaped vehicle lane markers.
29 Citations
22 Claims
-
1. A process for recognition of vehicle lane markings from image data, comprising:
-
evaluating morphological characteristics of vehicle lane markings using a priori knowledge, and using a matched-filter in order to extract image points, which are associated with vehicle lane markers, by measuring the average gray value of the background in the environment of the position to be examined, and evaluating an image point, which is potentially to be associated with the vehicle lane marking, on the basis of a comparison between background noise, the average gray value in the environment, and a gray value of the position to be examined. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21)
-
-
22. A process for recognition of vehicle lane markings from image data, comprising:
-
evaluating morphological characteristics of vehicle lane markings using a priori knowledge, using a matched-filter in order to extract image points, which are associated with vehicle lane markers, by measuring the average gray value of the background in the environment of the position to be examined, and evaluating an image point, which is potentially to be associated with the vehicle lane marking, on the basis of a comparison between background noise, the average gray value in the environment, and a gray value of the position to be examined, wherein after the extraction of the image points, which are to be associated with vehicle lane markings, these are digitized, wherein the intensities of the individual image points are compared with a threshold value, and the image points are only then drawn upon for further evaluation when their intensity exceeds this threshold, wherein the threshold value is determined from background noise using a threshold value regulator or controller, and wherein the threshold value regulator draws upon a priori knowledge regarding the expected surfaces occupied by the vehicle lane markings, which are directly correlated with the expected number of image points associated with the vehicle lane markings, and wherein the threshold value regulator or controller thereupon aims to supply the number of the image points extracted in the ROI preferably exactly to this expected value.
-
Specification