SYSTEM AND METHOD FOR RECOGNIZING A LANE

• US 20180260634A1
• Filed: 07/25/2017
• Published: 09/13/2018
• Est. Priority Date: 03/10/2017
• Status: Active Grant

• Current Assignee: HL Klemove Corporation
• Original Assignee: Mando-Hella Electronics Corp.
• Inventors: KIM, Shinwook
• Granted Patent: US 10,460,179 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 18/2163 Partitioning the feature space
• CPC Class Codes: G06F 18/217 Validation; Performance evaluation; Active pattern learning techniques
• CPC Class Codes: G06F 18/2413 based on distances to training or reference patterns
• CPC Class Codes: G06F 18/2414 Smoothing the distance, e.g. radial basis function networks [RBFN]
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06N 3/063 using electronic means
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06V 10/757 Matching configurations of points or features
• CPC Class Codes: G06V 10/764 using classification, e.g. of video objects
• CPC Class Codes: G06V 10/82 using neural networks
• CPC Class Codes: G06V 20/588 Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road

Abstract

A lane recognition system and method are disclosed. The lane recognition system according to an embodiment of the present invention includes an image sensor for capturing an image comprising a lane; a neuromorphic system in which a plurality of neurons in which a pattern vector related to a lane information are stored are connected by a parallel bus; and a lane recognition unit for generating [1×N] size input vectors of a predetermined size for a region of interest of the input image frame input from the image sensor, for inputting the generated vector to the neuromorphic system, and for recognizing a lane in the input image based on lane information of a neuron having a pattern vector most similar to an input vector among a plurality of neurons stored in the system.

Claims

1. Lane recognition system comprising:

an image sensor for capturing an image comprising a lane;

a neuromorphic system in which a plurality of neurons in which a pattern vector related to a lane information are stored are connected by a parallel bus; and

a lane recognition unit for generating [1×

N] size input vectors of a predetermined size for a region of interest of the input image frame input from the image sensor, for inputting the generated vector to the neuromorphic system, and for recognizing a lane in the input image based on lane information of a neuron having a pattern vector most similar to an input vector among a plurality of neurons stored in the system.

2. The system according to claim 1, wherein a pattern vector including vehicle information related to at least one of a lane color and a lane shape is stored in advance in the plurality of neurons.

3. The system according to claim 1, wherein the neuromorphic system simultaneously propagates the input vector to a plurality of neurons inside the neuron through the parallel bus, compares the input vector with the pattern vector stored in the plurality of neurons to obtain the neuron having the smallest relative distance value and outputs the category value of the selected neuron to the lane recognition unit;

and the lane recognition unit may recognize the lane in the input image based on the category values output from the neuromorphic system.

4. Lane recognition method comprising:

receiving an image frame comprising a lane;

generating a [1×

N] size window of a window of a predetermined size for an area of interest of the input image frame;

inputting the generated vector into the neuromorphic system in which the plurality of neurons storing pattern vectors related to lane information are connected by a parallel bus; and

recognizing a lane in the input image using lane information of the neuron having a pattern vector most similar to the input vector among the plurality of neurons stored in the neuromorphic system.

5. The system according to claim 4, wherein

propagating the input vector to the plurality of neurons simultaneously via the parallel bus in the lane recognition;

selecting a neuron having a smallest relative distance value by comparing the input vector with a pattern vector stored in the plurality of neurons;

outputting a category value of the selected neuron; and

recognizing a lane in the input image using the output category values.