Method and device for identifying target object in road

Abstract

The invention discloses a method and device for identifying a target object in a road, and relates to the technical field of computers. A specific embodiment of the method comprises the steps of collecting point cloud data of a road, wherein the point cloud data comprises three-dimensional space coordinate information and reflection intensity information of each point cloud; converting the point cloud data into four-dimensional semantic targets based on the three-dimensional space coordinate information and the reflection intensity information of each point cloud, and clustering the four-dimensional semantic targets to obtain a target point cloud cluster; and identifying a target object in each target point cloud cluster. According to the embodiment, sample learning and other data auxiliary registration are not needed, and the target object can be fully automatically and rapidly identified based on the spatial features of the three-dimensional point cloud data, and the hardware cost isreduced while the calculation process is simplified.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a method and device for identifying target objects in roads. Background technique [0002] In order to ensure the safety of vehicles, it is necessary to quickly extract and locate target objects on the road, such as manhole covers, traffic signs, etc., to provide decision-making for obstacle avoidance or passage for the vehicle end. At present, there are mainly the following technical means for target object extraction and positioning: [0003] (1) Based on a large number of positive and negative samples, machine learning or deep learning is used for model training, and prediction and recognition are performed on the basis of learning and training models; [0004] (2) Relying on image data and point cloud data at the same time, the image and point cloud data are registered according to the strict conversion relationship, and the features of the two are combined for identificat...

AI Technical Summary

Problems solved by technology

[0006] (1) The method of learning and predicting based on samples requires a lot of manual labeling of positive and negative samples in the early stage; the accuracy of extraction and positioning is limited by the number of positive and negative samples and the scenes covered by the samples, small data samples or unbalanced positive and negative samples will lead to a decrease in accuracy

[0007] (2) The extraction and positioning method of fused image and point cloud data features needs to accurately calibrate the external parameter conversion relationship between the sensors of the acquisition equipment, and the calculation process is complicated; to obtain image data, camera sensors need to be installed, which increases hardware costs

Images