And constructing target distance detection method of four-channel virtual image by using multi-source information

Abstract

The invention relates to a target distance detection method for constructing a four-channel virtual image by using multi-source information, which comprises the following steps: acquiring original point cloud data by using a millimeter-wave radar for information processing, and determining radar original point information belonging to the same target to obtain a target size and a target reflectioncenter position; according to a reflection center position of a target under a radar plane and a target center pixel position in an image acquired by a monocular camera, searching a space conversionrelation of the two sensors through a joint calibration method, and simultaneously realizing association of asynchronous heterogeneous multi-source information in combination with time synchronization; constructing a virtual four-channel picture containing distance information according to the incidence relation between the millimeter-wave radar and the image data; and building a convolutional neural network according to the virtual four-channel picture to realize target detection. The method can improve the distance prediction capability of target detection, achieves the light weight of a network structure, saves the calculation resources, and improves the spatial information prediction precision and speed of a conventional visual 3D target detection algorithm.

Description

technical field [0001] The invention relates to the field of environment perception of smart cars, in particular to a target distance detection method using multi-source information to construct a four-channel virtual image. Background technique [0002] Accurate, reliable, and robust environmental perception is critical in smart vehicle systems. Image information contains rich semantic features. With the development of artificial intelligence and deep learning, the realization of vision-based target detection algorithms based on convolutional neural networks has become increasingly mature and has become a hot research topic. However, since monocular vision cannot directly obtain target distance information and convolutional neural networks are more suitable for classification tasks, vision-based target distance detection still needs to be improved: existing 3D visual target detection algorithms generally cannot achieve distance detection in driving tasks Accuracy requireme...

AI Technical Summary

Problems solved by technology

The former is interfered by the clutter detected by the radar, and if the radar misses detection, it will affect the target detection rate; the latter algorithm is time-consuming, and at the same time, it fails to make full use of the perception information provided by the millimeter-wave radar to achieve more effective fusion

Images