论文部分内容阅读
用单摄像机所获取的二维(2D)图像来估计两坐标之间的相对位姿和运动在实际应用中是可取的,其难点是从物体的三维(3D)特征投影到2D图像特征的过程是一个非线性变换,把基于单目视觉的位姿和运动估计系统定义为一个非线性随机模型,分别以迭代扩展卡尔曼滤波器(IEKF)、一阶斯梯林插值滤波器(DD1)和二阶斯梯林插值滤波器(DD2)作非线性状态估计器来估计位姿和运动.为了验证每种估计器的相对优点,用文中所提方法对每种估计器都作了仿真实验,实验结果表明DD1和DD2滤波器的特性要比IEKF好.
Estimating the relative pose and motion between two coordinates using a two-dimensional (2D) image acquired by a single camera is desirable in practical applications with the difficulty of projecting from the three-dimensional (3D) feature of the object to the 2D image feature Is a non-linear transformation. The pose and motion estimation system based on monocular vision is defined as a non-linear stochastic model. The iterative extended Kalman filter (IEKF), the first-order Strait interpolation filter (DD1) Second Order Steinlin Interpolation Filter (DD2) is used as a nonlinear state estimator to estimate pose and motion.In order to verify the relative merits of each estimator, we simulate each estimator using the proposed method, Experimental results show that the characteristics of DD1 and DD2 filter better than IEKF.