Super-pixel based road image segmentation method

Abstract

The invention relates to a super-pixel based road image segmentation method, which belongs to the field of image processing and is mainly applied to automobile road driving. The road image segmentation method comprises the steps of performing over-segmentation on a road image by using SLIC super-pixels, proposing a method for detecting an under segmentation region based on the regional color maximum difference in allusion to a circumstance that the SLIC super-pixels are easy to have an under segmentation phenomenon in a thin strip region, and adding a new clustering center to correct the undersegmentation; then performing regional feature extraction based on Gabor filtering and an LAB color space; and performing re-fusion on the over-segmentation region through calculating the similaritybetween the regions on the basis of considering the spatial position adjacency of the regions, and achieving accurate segmentation for a road region under a complex environment. The super-pixel basedroad image segmentation method serves as an important basic link of a vehicle-mounted advanced driver assistance system, and has a good segmentation effect and real-time processing ability for complexurban environment road images.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a road image segmentation method based on superpixels. Background technique [0002] The issue of road traffic safety has always attracted much attention, and it has become an inevitable trend to install intelligent vehicle assistance systems in vehicles. The intelligent driving assistance system is mainly used to perceive the surrounding driving environment, such as drivable road area, surrounding obstacles, traffic sign information, etc., so as to provide assistance and warning to the driver. At present, there are many related researches on obstacle detection, traffic light recognition, road detection and so on. Among them, the quality of road image segmentation, the accuracy of boundary and lane line positioning will directly affect the safe driving of vehicles, and it is also an important technical link in subsequent image processing, analysis, and understanding. ...

AI Technical Summary

Problems solved by technology

[0007] In order to solve the accuracy and speed problems of existing road image segmentation methods, the present invention proposes a road image segmentation method based on superpixels. The method first utilizes the SLIC superpixel segmentation method to over-segment the color road image, and then extracts the road image based on superpixels. Image color and texture features, finally, combine the superpixel color and texture features to merge similar adjacent superpixels in the road image to obtain the road and lane line area of ​​the image

Images