Pseudo point cloud data generation method of laser radar

Abstract

The invention relates to a pseudo point cloud data generation method of a laser radar. The method comprises the steps of by a processor, selecting a first color image, a second color image and a thirdcolor image, based on time data, obtaining two frames of sparse depth images before and after the time data corresponding to the second color image, namely a first sparse depth image and a second sparse depth image, determining an optical flow according to the offset displacement of the selected first pixel point, obtaining an intermediate frame sparse depth image according to the optical flow, the first sparse depth image and the second sparse depth image, taking the first sparse depth image, the second sparse depth image and the intermediate frame sparse depth image as inputs of a convolutional neural network, deducing and outputting an intermediate frame dense depth image, taking the second color image and the intermediate frame dense depth image as the input of a U-net network, deducing and outputting an intermediate frame fine depth image, and performing coordinate conversion processing according to pixel points in the intermediate frame fine depth image to obtain pseudo point cloud data of the intermediate frame fine depth image.

Description

technical field [0001] The invention relates to the technical field of laser radar, in particular to a method for generating pseudo point cloud data of laser radar. Background technique [0002] In recent years, the unmanned vehicle market has been booming. After Google, mainstream unmanned vehicle R&D teams such as Baidu and Uber all use lidar as one of the most important sensors, and use the combination of lidar and camera to realize Accurate judgment of obstacles and other information during unmanned driving. [0003] However, lidar has the following defects: (1) the data collected by lidar is very sparse, and only about 4% of the images in the 2D depth map obtained by projection of the collected point cloud data have pixel values; The scanning frequency is very low, usually about 10Hz for lidar sensors. At present, the highest scanning frequency on the market is Velodyne's 64-line 3D lidar, and its maximum scanning frequency can only reach 15Hz. [0004] Usually, the ...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to address the defects of the prior art, to provide a pseudo point cloud data generation method for laser radar, to solve the problem that the scanning frequency of laser radar is too low, and the point cloud data generated by scanning is coarse in time and space and inaccurate question

Images