High-resolution radar imaging system is widely used in both military and civil fields. Wide-band radar signals are necessary in high-resolution radar imaging, while these signals introduce high-sampling rate, high-transmission rate, large data storage and high difficulties to real time signal processing in radar system. These problems are resulted from Nyquist sampling theorem which requires the sampling rate to be more than?two times of the?bandwidth of the signal. As a consequence, searching for new signal processing and data acquisition methods is in urgent requirement. When dealing with some signals with the property of sparsity, the theory of compressed sensing which is different from the?Nyquist sampling?theorem, gets a group of numerical values through noncorrelation-measurement and the number of these measurements is much less than that of points sampled according to Nyquist sampling theorem. Then, we can reconstruct original-signal accurately by reconstruction-algorithms. The premise of using the theory of compressed sensing is the signal’s sparse property and the key to the theorem is the relationship between the measurement matrix and sparse degree. Meanwhile, the important support is the reconstruction algorithm. As we all know, response function of?radar observations of the?scene is usually?sparse?and this property leads to the?sparsity of wide-band radar echo in some?form. Based on this property, the application of?the theory of compressed sensing in radar signal processing becomes possible. In this paper, the principle of signal sampling and reconstructing according to the theory of compressed sensing has been introduced briefly and a sparse matrix structure scheme for linear frequency modulation signal （LFM） commonly used in radar is proposed with the help of the emission signal. At the same time, the LFM signal model is structured with the help of MATLAB. Then, the processes of sampling and reconstruction of LFM and 2D imaging with the theory of compressed sensing are also simulated. Besides, this paper also studied the?different reconstruction?algorithms and makes the algorithm?effectiveness comparison. The results show that during the process of broadband radar echo signal processing, the theory of compressed sensing can effectively relieve the pressure on the echo data storage and transmission through reducing sampling rate. This advantage can be widely used in wide-band radar target detection.