Radar altimeter is an indispensable airborne electronic equipment in many aircraft. It has been widely used in various military and civilian equipment. The radar altimeter based on FM continuous wave constant beat frequency system has the advantages of strong anti-interference and no distance blind area, and has a good application prospect. However, since the radar altimeter's target is the ground, and the ground is a complex surface target, the echo information reflected from the ground is very rich, and it is difficult to accurately extract the height information. Therefore, to solve the above problems, this paper applies convolutional neural network to the spectrum frontier detection of altimeter radar, and proposes a spectrum frontier detection algorithm based on VGG16-Net.