The invention discloses an automobile fatigue driving detection method based on multi-source information fusion. The automobile fatigue driving detection method comprises the steps that the physiological signal change of a driver is detected through a Doppler radar detection unit in a cab, and physiological signals are obtained and stored in real time; the change amount of steering wheel angle information in preset time is obtained, and the physiological signals obtained in a current continuous time period are enabled for data processing according to the change amount; whether fatigue driving occurs or not is judged according to the physiological signal change. Compared with the prior art, a Doppler measuring technique is adopted for the first time to monitor the fatigue degree of a driver, and meanwhile the driving measurement accuracy is improved by detecting the steering wheel angle information and through Doppler information processing. In addition, a general Doppler radar sensor works in a continuous wave mode, a multistage filter circuit is correspondingly designed, accordingly non-contact detection of physiological signals of a human body is achieved, and the signals serve as a basis to judge fatigue driving.