In order to improve autonomous velocity measurement performance for UAV (unmanned aerial vehicle) low-altitude flight, a autonomous velocity measurement method for UAV based on photoelectric image feature is proposed. Firstly, two frames of continuous photoelectric image that have overlapped scene are acquired used photoelectric equipment. The method detects point features and line features from the before and after images respectively, and judges the extracted image features, and extract the valid image features. Then, the extracted image features from the before and after images is associated. According to the pixel position of association image features, the method calculates pixel displacement vector, and exclude the pixel displacement vector with large deviation. Finally, according to the parameter of photoelectric equipment and feature of displacement vector, the method calculates the ground-speed of UAV. Using the UAV with photoelectric equipment flight test, the experimental results show that the method can effectively reduce computational complexity and have better real-time performance, and meet the autonomous velocity measurement requirement for UAV and other low-altitude flight platforms.