Lane line detection method for assisting driving

The invention discloses a lane line detection method for assisting driving, and the method comprises the steps: carrying out the training of an image sample based on a multi-task convolutional neuralnetwork, assisting the recognition of a lane line through the correlation between a vehicle and a lane line, improving the detection safety of the complex and variable automatic driving of an actual driving environment, and reducing the hardware computing power processing requirement. The lane line detection method not only enables auxiliary driving to be safer, but also has the advantages of being higher in working efficiency, more accurate in recognition precision, capable of reducing misjudgment and the like. 本发明公开了一种用于辅助驾驶的车道线检测方法，通过基于多任务卷积神经网络对图像样本训练，利用车辆与车道线的关联性来辅助车道线的识别，提高对实际行车环境复杂多变自动驾驶的检测安全性，降低硬件算力处理要求；本发明的车道线检测方法不仅使辅助驾驶更安全，还具有工作效率更高、识别精度更准、减少误判等优点。