Driving distance detection is one of the key technologies of the car''s active safety driving assistance system. This paper proposes a method of real-time monocular depth estimation, to improve the accuracy and real-time performance of the distance detection during the vehicle driving process. First, the distortion model is constructed, and the camera calibration algorithm is used to calibrate the monocular camera. Then, the license plate is used as the target location benchmark of the front vehicle, and the license plate information of the front vehicle is extracted quickly by using the license plate filtering algorithm of color and contour. Finally, based on directional gradient histogram feature and support vector machine, the license plate is located accurately. Experimental results show that, compared with other methods, the perspective N-point depth estimation model fused with known license plates has high accuracy and good real-time performance. This method has a recognition rate of 99.326% for the location of the license plate of the front vehicle, and the detection error of the driving distance is less than 10%, and the time required to process an image is about 170 ms, which meets the application requirements for the detection of the distance between vehicles during the vehicle driving process.