The invention specifically relates to a velocity measurement method for a visual odometer based on multi-field-depth information, belonging to the technical field of visual navigation. The velocity measurement method provided by the invention comprises the following steps: 1, selecting a proper binocular camera according to task characteristics; 2, calibrating installation errors of a MEMS inertial navigation/the visual odometer; 3, extracting characteristic points of an image based on gray-variance information; 4, measuring distance between the characteristic points by using the binocular camera; 5, identifying characteristic points in images of front and rear frames so as to realize velocity measurement; and 6, eliminating velocity measurement noise by using a wavelet analysis method. The method provided by the invention overcomes the technical problems that conventional velocity measurement methods for the visual odometer have low precision and are hard to apply to intelligent vehicles needing high-precision long-endurance navigation. According to the method provided by the invention, MEMS inertial navigation auxiliary speed information added into velocity measurement and calculation of the visual odometer can substantially improve velocity measurement precision, so complexity of calculation is reduced and the method is applicable to long-endurance navigation.