Yuehua Cheng

Research on computer vision-based for UAV autonomous landing on a ship

In this paper, a novel approach to UAVs automatic landing on the ships deck is proposed. We present the design of the cooperative object, and then begin our basic research on UAV autonomous landing on a ship by using computer vision and affine moment invariants. We analyze the infrared radiation images in our experiments by extracting the target from the background and then recognizing it. Also, we calculate the angle of yaw. We study the basic research concerning automatic UAV navigation and landing on the deck. Based on our experiments, the average recognition time is 17.2ms which is obtained through the use of affine moment invariants. This type of speed is expected to improve the reliability and real-time performance of autonomous UAV landing.

Global motion estimation based on kalman predictor

In order to detect moving object by a rotated camera in video surveillance, block-based motion estimations (BME) are performed first and global motion parameters are estimated. A novel search algorithm that based on kalman filter is proposed. The algorithm is a kind of block-matching motion estimation algorithm. First feature points are extracted from current frame and then feature points are used as the central points in block matching between consecutive frames, then the 3σ rule is used to remove blocks of error. Kalman filter is used to search matching blocks and results have shown that a total decrease by about 95% in computation time is achieved compared to the classical full-search BME process in global motion estimation.

A simple, robust and fast method for the perspective-n-point Problem

Abstract In this work, we present a simple, robust and fast method to the perspective-n-point (PnP) problem for determining the position and orientation of a calibrated camera from known reference points. Our method transfers the pose estimation problem into an optimal problem, and only needs to solve a seventh-order and a fourth-order univariate polynomial, respectively, which makes the processes more easily understood and significantly improves the performance. Additionally, the number of solutions of the proposed method is substantially smaller than existing methods. Experiment results show that the proposed method can stably handle all 3D point configurations, including the ordinary 3D case, the quasi-singular case, and the planar case, and it offers accuracy comparable or better than that of the state-of-art methods, but at much lower computational cost.