Algorithm And Application About Three-dimensional Imaging Laser Radar Based OnCompressive Sensing

Compressed sensing (CS) theory is a booming theory in recent years, it can reconstruct the signal accurately with far less than the Nyquist sampling frequency while there's no loss of information in the meantime. This method takes a small amount of sampling data to simultaneously realize collection and compression, and significantly reduces the demand of system for data transmission and storage, then saves data processing time and cost of the system hardware.Compressed sensing theory has three key elements:(1) Signal sparse transformation (2) Design of observation matrix (3) signal reconstruction. The main part of this study is the algorithm of signal reconstruction and two different kinds of CS laser radar applications. As the key procedure in compressive sensing theory framework, the selection of algorithm directly impacts the efficiency and accuracy of the reconstructed signal, and the possibility in the practical application. After lots of research and summary, signal reconstruction algorithm can be classified into four categories, namely the convex optimization algorithm, greedy algorithm, statistical optimization method and combination method. Four kinds of algorithm in this paper have their own characteristics. We focus on the several typical algorithms such as gradient projection (GPSR), iterative threshold (1ST) and orthogonal matching pursuit (OMP) and so on. The research on these basic algorithms has huge guiding significance on the improved algorithms put forward later.Contraposing the limitation of the OMP algorithm, we propose a optimized OMP algorithm. Firstly making a wavelet transform on the 2-D image data and we obtain the subimage which contains different feature information, and we employ the compressive sampling OMP algorithm on subimage respectively. At last we use the wavelet inverse transformation to recover the original image. The advantage of this approach is that it can eliminate the interference between each other, thus to get better results.This article also designed two kinds of compressive sensing laser radar. The first radar is based on the DMD micro-mirror to achieve compression sampling. The micro-mirror different flip states simulate the measurement matrix values so as to realize the modulation of the target object optical path. The data processed by the optimized OMP algorithm. Another is full-waveform laser radar, the application of compression sensing not only can greatly reduce the amount of data acquisition, but also don't require expensive high speed ADC, which greatly decreases the cost of experiments. The system consists of a single photodiode and multi-channel gate integral circuit to achieve compression sampling. We control the switch state of integral circuit according frequency doubling by FPGA to simulate measurement matrix-Toeplitz matrices. The target locations can be directly recovered by deconvolution, and also achieve the full waveform data.