As an important sensor for the environmental perception functions of the Advanced Driver Assistant System (ADAS), automotive radar is used for targets detection and parameters (range, velocity and angle）measurement . The procedure of constant false alarm rate (CFAR) plays an important role in adaptive targets detection in a noise or clutter environment. However，the traditional cell averaging ( CA-CFAR ) algorithm has a masking effect in the all multiple target situations and the ordered statistics (OS-CFAR) algorithm needs a large amount of computation. These one-dimension algorithms are not stable separately in the on-road environmental perception. Therefore, this paper focuses on the multiple target on-road environment, and the LFMCW radar simulation system is built to produce a series of rapid chirp signals, which can be used to measure the target parameters (distance, radial velocity, angle). Then the echo signals are converted into a two-dimensional range-Doppler matrix (RDM) through twice fast Fourier transform (FFT). Through the processing of the two-dimensional matrix, a fast and effective two-dimensional CFAR algorithm is proposed. The experimental results demonstrate that the new two-dimensional algorithm can enhance the accuracy and robustness of the multi-target detection in the actual road environment.