为了解决传统目标检测算法在非均匀噪声环境下检测性能严重下降的问题，提出了一种自由滑窗式恒虚警检测算法，在被检测单元两侧各取 N－1个参考单元作为初始自由参考窗，然后从左向右依次滑动 N次，得到 N个滑窗，根据滑窗噪声功率的平均值与被检测参考单元功率值的比较结果，选择相应的滑窗噪声功率平均值集合取均值，再乘以参数因子 T得到比较门限阈值 S，根据被检测单元的功率与 S进行比较，确定是否为有效目标。经过与其他算法进行仿真对比，该算法具有最优的检测性能，检测率 98.93%，误检率 2.28%，并成功应用于车辆开门预警系统，经测试，预警率大于 98.10%，虚警率小于 2.80%。结果表明，该算法提高了非均匀噪声环境下目标的检测概率，具有良好的检测性能。

In order to solve the problem that the detection performance degradation of the conventional target detection methods in non-homogenous environments, a sliding window-constant false alarm rate (SW-CFAR) detection algorithm was proposed. Free units on the both sides of the reference cell under test were set as an initial reference window which was slid N times from left to right to get N sliding windows. Based on the comparison results between the average power value of the sliding window and the power of cell under test, the proper average power value of the sliding windows were selected to obtain free sliding window average power which was multiplied by the factor T parameters to get comparison threshold S. Power comparison results of the cell under test and detection threshold S determined whether there was target or not. According to simulation and analysis comparison results with other algorithms, the proposed algorithm had the best detection performance, whose detection rate was up to 98.93% and false detection rate was down to 2.28%. The algorithm was successfully applied to a vehicle door open warning system, the average early warning rate was up to 98.10% and false alarm rate was down to 2.80%. The experimen-tal results show that the proposed algorithm improves the target detection probability and has a better de-tection performance in non-homogenous environments.