In the field of civil aircraft flight test, the speed of aircraft is one of the most important parameters. Generally, there are different methods for the measurement of aircraft speed, including differential GPS measurement, photoelectric theodolite measurement, etc. However, each of them has its own limitations. Commonly, multiple speed measurement methods are used simultaneously to verify each other in important flight test mission. In order to enrich the aircraft speed measurement methods, this paper proposes an aircraft speed measurement method based on computer vision of high-speed camera. The method utilizes the similarity transform model in computer vision theory, and can measures the speed of the aircraft in the world coordinate system. Then, an analysis method of theoretical speed measurement accuracy based on the speed measurement method is proposed by the paper. Finally, in the actual flight test, the method is used to measure the heading speed of the aircraft. Comparing the measurement results of the differential GPS system and the measurement method proposed in this paper, which verifies the accuracy and of the method. The method can be applied not only in the field of flight test but also in a wider range of engineering applications.