Low-altitude small target detection method based on R-D graph and deep neural network

Abstract

The invention discloses a low-altitude small target detection method based on an R-D graph and a deep neural network. The method comprises the following steps: step 1, preprocessing an echo signal acquired by a radar receiver to obtain an R-D graph of the signal; 2, constructing a deep convolutional neural network, and predicting the category probability that each local region in the R-D graph belongs to a target and a background and the offset between the center of the region and the target position when the region contains the target by using the network; and 3, judging whether a target exists or not and judging the target position when the target exists in combination with the predicted category probability and the offset of position regression. According to the method, regression of the target position is added on the basis of category prediction, and the accurate position of the target is obtained through voting statistics of the regression result, so that the radar small target detection effect with higher prediction accuracy and lower false alarm rate can be realized.

Description

technical field [0001] The invention belongs to the field of radar signal processing and image recognition, and relates to a low-altitude small target detection method based on an R-D (Range-Doppler) map and a deep neural network. Background technique [0002] In recent years, in response to the needs of the military and civilian fields, low-altitude small aircraft represented by drones have developed rapidly. At the same time, this kind of target has the characteristics of "small scattering cross-sectional area, slow flying speed, and low flying altitude", which makes it difficult to effectively use the traditional frequency domain filtering detection technology. In order to reduce the adverse effects of clutter and interference, the traditional radar target detection method usually adopts the method of constant false alarm rate (CFAR) to identify the target after data preprocessing, and the identification process has the following two problems: [0003] 1. The probability...

AI Technical Summary

Problems solved by technology

[0003] 1. The probability distribution of receiving electromagnetic wave signals is difficult to predict, and CFAR processing cannot accurately make a judgment on the detection unit according to the energy distribution of the reference unit;

[0004] 2. CFAR processing relies on manual adjustment of the threshold, which is weakly adaptive and cannot actively adjust with signal changes

Images