Denoising method for a lidar bathymetry system based on a low-rank recovery of non-local data structures

Abstract

The lidar bathymetry system (LBS) echo is often contaminated by mixed noise, which severely affects the accuracy of measuring sea depth. The denoising algorithm based on a single echo cannot deal with the decline of the signal-to-noise ratio and impulse noise caused by sea waves and abrupt terrain changes. Therefore, we propose a new denoising method for LBS based on non-local structure extraction and the low-rank recovery model. First, the high-frequency noise is eliminated based on the multiple echo in a small neighborhood, and then the matrix is constructed based on the processing results in a larger range. Then, we make full use of the structural similarity between LBS echoes by transforming the echo denoising issues into low-rank matrix restoration to further eliminate the noise. The experimental results show that this method can effectively preserve the seafloor signal and eliminate the mixed noise.

Coupling Dilated Encoder–Decoder Network for Multi-Channel Airborne LiDAR Bathymetry Full-Waveform Denoising

“…The Richardson-Lucy deconvolution [13,25] algorithm can be used to extract the interface response function to separate multiple targets with a short distance, but the algorithm relies on multiple iterations, which tend to produce noise and sound effects. Therefore, superimposing structurally similar subsequences in multiple frames [19] can significantly improve the peak signal-to-noise ratio of submarine signals. B-spline [26] and Wiener deconvolution algorithms [27] require no prior knowledge and avoid short-distance stacking losses, but they are too sensitive to noise.…”

“…Although multi-channel information has been proven to be effective for echo classification, denoising and signal enhancement have not been used [5,6]. Due to the structural similarity between consecutive waveforms and the physical correlation between different channels of the same frame [18], they can be used to further improve the signal sampling rate and reduce noise [19]. However, with the increase in the detection distance, the intensity of the LiDAR echo signal steadily decreases.…”

A Classification Method for Airborne Full-Waveform LiDAR Systems Based on a Gramian Angular Field and Convolution Neural Networks

“…Due to many factors, such as terrain fluctuation, noise, and multiple diffuse reflections of objects, it is easy to miss some overlapping echoes, which leads to a reduction in classification accuracy. To improve the generalization performance and robustness of the convolutional neural network, we refer to the contrast experiment of airborne echo denoising [29] and increase the diversity of samples by adding abrupt noise and Gaussian white noise to the original noise.…”

Calibration on X-band radar ranging with laser lidar

“…Although based on the same principle as radar, lidar is a ranging device that uses light rather than radio waves [10,11]. By using light of a short wavelength with short pulse, high accurate measurements can be obtained using a precise-directional and narrow beam [12]. So some airborne [13,14] and satellitebased [15] remote sensing use the lidars.…”