In order to solve the problem of cloud cover over the heliostat field of tower solar power station, a velocity measurement method was proposed which was used to forecast the moving cloud over the heliostat field. Firstly, the cloud image was captured by a fixed camera, then the images with different frames were extracted by software, and those operations were performed, such as gamma transformation, image segmentation based on genetic algorithm, cloud target detection, and motion cloud layer matching. Finally, the movement speed and direction of cloud layer was calculated according to the pixel coordinates obtained by the matching. The experimental results show that compared with the data of optical flow method, the image segmentation method requires less computation and the location prediction error is small. It is concluded that the image segmentation method for cloud velocity measurement is feasible, which can provide theoretical basis for cloud cover prediction in the cloud monitoring of tower solar power plant.