Lane Detection (Part I): Mono-Vision Based Method

Abstract : In this report, we introduce a mono-vision based lane detection method (pure mono-vision without the aide of any other sensor). The lane detection method consists of the sub-parts of lane mark segment extraction, lane model fitting, and lane tracking. The lane mark segment extractor is based on some criteria that are only related to the inherent physical properties of the lane marks. The lane model is parameterized in the world (the ground plane) reference instead of in the image reference. The lane tracking is carried out in the framework of the particle filter, including the evolution and the update of the lane state; a mechanism that executes on-line estimation of the vehicle pitch angle is naturally incorporated in the lane tracking process. Read-data tests are given to demonstrate the performance of the mono-vision based lane detection method.
Document type :
Reports
Complete list of metadatas

Cited literature [18 references]  Display  Hide  Download

https://hal.inria.fr/hal-00781274
Contributor : Hao Li <>
Submitted on : Friday, January 25, 2013 - 6:15:38 PM
Last modification on : Thursday, August 2, 2018 - 12:02:03 PM
Long-term archiving on : Friday, April 26, 2013 - 3:58:17 AM

File

RT-433.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00781274, version 1

Collections

Citation

Hao Li, Fawzi Nashashibi. Lane Detection (Part I): Mono-Vision Based Method. [Technical Report] RT-0433, INRIA. 2013. ⟨hal-00781274⟩

Share

Metrics

Record views

739

Files downloads

2979