An unmanned lane line detection method based on a generative adversarial network

Abstract

The invention discloses a driverless lane line detection method based on a generative adversarial network. The driverless lane line detection method comprises three parts of a lane line picture generation network, a discrimination network and a detection algorithm, and is characterized by firstly obtaining the training data of the generation network and a discrimination network, then inputting thelow-resolution image into the generation network to generate a super-resolution image; inputting the high-resolution image and the super-resolution image into a judgment network for definition judgment; and capturing data distribution of the super-resolution image and the high-resolution image according to a judgment result, performing the adversarial training based on the data distribution by the generation network and the judgment network until Nash equilibrium is reached to obtain an optimal generation network, and inputting the super-resolution image generated by the optimal generation network into a detection algorithm to perform lane line identification. The method adopts the generative adversarial network to carry out the driverless lane line detection, and can effectively improvethe lane line detection accuracy.

Description

technical field [0001] The invention relates to an unmanned driving lane line detection method with enhanced image resolution based on a generative confrontation network, which belongs to computer graphics processing technology and artificial intelligence technology. Background technique [0002] Although the traditional lane line detection algorithm can cope with most situations, if it encounters extreme weather conditions such as fog and rain, the camera will be affected, so that the photos it can get will become blurred, resulting in unmanned driving. Safety risks are constantly accumulating, and once an accident occurs, it is easy to bring danger to the lives of passengers. The research on generative confrontation network can greatly reduce the safety problems of unmanned driving, make unmanned driving technology more trustworthy by the public, promote the vigorous development of unmanned driving technology, and bring vital benefits to automobile companies and ordinary u...

AI Technical Summary

Problems solved by technology

Including the traditionally used bicubic image interpolation algorithm, but it generally deals with smaller images. Once the magnification of the image is more than 4 times, it is easy to make the result appear too smooth, and lack the realism of some details.

Images