Echo feature extraction method for single-pulse laser radar based on deep learning

Abstract

The invention discloses an echo feature extraction method for single-pulse laser radar based on deep learning. The method comprises the steps of collecting various laser radar echo signals with complete signal features, carrying out preprocessing on the effective parts of the collected echo signals, and carrying out splicing, forming training data by theobtained spliced sequence with echo feature information, carrying out waveform recovery on distorted echo signals based on an LSTM-RNN model, and particularly carrying out calculation prediction on a saturated part of saturated distortion so as to obtain part of echo information lost due to waveform distortion, and then carrying out feature extraction through a convolutional neural network to obtain the waveform features of radar echoes. The method is applied to analysis of single-pulse laser radar echo signals and does not completely depend on hardware equipment, so that feature information of the distorted waveform signals can be extracted from the distorted waveform signals.

Description

technical field [0001] The invention relates to waveform feature extraction technology, in particular to a method for extracting echo features of single-pulse lidar based on deep learning. Background technique [0002] Lidar is the product of the combination of radar principle and laser technology. Using laser as the detection beam, it has a series of advantages such as long range, high sensitivity, high spatial resolution, and strong anti-interference ability. It has broad application prospects in the fields of environment perception and so on. [0003] Signal processing is the most important part of the lidar detection system. Its purpose is to accurately analyze, diagnose, compress and quantify the echo signal, and quickly realize the transmission, storage and accurate reconstruction of the signal. Due to the influence of the complex and changeable detection environment, the echo signal of the lidar not only carries the information components of the target, but also is m...

AI Technical Summary

Problems solved by technology

Solving the saturation problem will be of great help to the subsequent feature extraction. Generally speaking, the traditional laser radar ranging system will discard the saturated signal, but this will have a certain impact on the performance of the radar, so it is necessary to use it as much as possible. Avoid the occurrence of saturation, or reduce the influence of saturated signals on the feature extraction of lidar echoes

Images