针对距离估计问题，提出了一种新颖的利用惯性传感器和单目摄像机的距离估计系统，该系统通过结合惯性导航算法和图像处理算法来实现距离估计。首先通过移动单目摄像机，得到一个目标的两幅图像，并利用改进后的Criminis算法提取两个图像间的特征值，再利用惯性导航算法对摄像机的运动进行估计运算，计算出摄像机移动的基线估计，最后通过最小二乘法得到估计距离。同时为了减少误差，在摄像机移动的过程中利用了零速度更新。通过实际环境实验，我们对该算法的准确度进行了评估，在几米范围内，它的平均距离误差是3.9%。%To estimate the distance between the target and moving camera,a novel distance estimation system has been proposed using a inertial sensor and a monocular camera system.This system realizes the distance estimation through a combination of inertial navigation algorithm and image processing algorithm. First of all,by moving a single camera, two images of a target are got,and the improved Criminis algorithm can be used to extract the two image feature value. The inertial navigation algorithm is used for camera motion estimation and calculated the baseline estimate of the cam-era movement. The final estimated distance is obtained by least square method. In order to reduce the error, zero velocity update is used in the process of the camera movement. Through the actual environment experiment,we have to evaluate the accuracy of the algorithm confined within a few meters,its average distance error is 3.9%.
展开▼
