Image semantic reconstruction method based on laser radar data

The invention discloses an image semantic reconstruction method based on laser radar data, and relates to the technical field of traffic safety. The method comprises the following steps: collecting streetscape image data, and obtaining a semantic segmentation image of an image; taking a laser radar map as the input of a deep convolutional neural network, taking the obtained semantic segmentation map as the label of the deep convolutional neural network, and taking the minimum error, i.e., the minimum loss function, between the label and the network output as a target training network model; and collecting the laser radar map and inputting the laser radar map into the trained deep convolutional neural network model. A fitting relationship between laser radar data and image semantic data isobtained; when the visual image and the image semantic signal cannot be acquired, the laser radar data is utilized to predict the image semantic data of the corresponding visual scene and reconstructthe semantic information o