An Image Orientation Estimation Method Based on Gradient Histogram and Point Matching

In recent years, the application of UAV has been more and more widespread. Attitude estimation as the main research direction of the control system has attracted much attention from industrial and academic area. The performance of traditional sensor fusion method is degraded significantly when the system is placed in high noise environments. Vision based estimation method, which depends on image geometry, has high robustness in complex environment. In this paper, we propose an image orientation estimation method based on gradient orientation histogram and point matching. In order to illustrate the performance of the proposed method, several simulations have been carried out. The first experiment is orientation estimation based on different point matching methods. The results are feasible, but contain a big error gap between different feature descriptor methods. The real-time algorithm has a large error because of the bad matching samples. Therefore a measurement method based on gradient orientation histogram is proposed to correct the results of real-time point matching orientation estimation method. This method use the output of gradient orientation histogram filter the results of pure point matching orientation estimation. The experiment results show that the proposed method can effectively shrink the error of real-time point matching orientation estimation.