| 刘佳,闫冬,郭斌.基于改进卡尔曼滤波的增强现实跟踪注册算法[J].高技术通讯(中文),2020,30(12):1225~1233 |
| 基于改进卡尔曼滤波的增强现实跟踪注册算法 |
|
| |
| DOI：10.3772/j.issn.1002-0470.2020.12.003 |
| 中文关键词: 卡尔曼滤波； FAST算法； 加速鲁棒特征(SURF)算法； 增强现实（AR）； 姿态估计 |
| 英文关键词: Kalman filtering, features from accelerated segment test (FAST) algorithm, speeded-up robust features (SURF) algorithm, augmented reality (AR), attitude estimation |
| 基金项目: |
|
| 摘要点击次数: 1887 |
| 全文下载次数: 1415 |
| 中文摘要: |
| 针对传统应用在特征点匹配技术方向的增强现实系统部分存在鲁棒性不足和实时性较差问题，提出一种新的区域跟踪匹配算法。首先利用改进的FAST算法在不同尺度上提取像素信息建立低维模板进行快速定位。其次在跟踪阶段使用加速鲁棒特征(SURF)检测关键点在线实时匹配的同时，借助改进的非线性化的卡尔曼滤波算法预测运动轨迹，缩减匹配区域。接着对使用上述的算法截取部分图像帧，利用上一帧的观测值对下一帧进行预测并估计相机姿态，同时使用随机抽样一致(RANSAC)算法剔除离群值。最后通过得到的相机位姿矩阵进行虚拟物体的注册叠加。实验结果表明，本文提出算法在综合评价中占据优势，在光照变换、旋转变换、尺度变换方面都具有较好的鲁棒性，并具有优越的实时性能。 |
| 英文摘要: |
| In order to improve robustness and real-time performance of traditional augmented reality systems applied in feature point matching, a new region tracking matching has been proposed. Firstly, features from accelerated segment test (FAST) algorithm is used to extract pixel information on different scales to establish a low-dimensional template for fast positioning. Secondly, in the tracking stage, speeded-up robust features (SURF) algorithm is used to detect the key points on-line and real-time matching. At the same time, the improved non-linear Kalman filter algorithm is used to predict the motion trajectory and reduce the matching area. Part of the image frame from the above algorithms is intercepted, and the observed value of the previous frame is used to predict the next frame and estimate the camera attitude. At the same time, random sample consensus (RANSAC) algorithm is used to eliminate outliers. Finally, the registration of virtual objects is superimposed through the camera position matrix gained. In this paper, the algorithm is equipped with advantages in comprehensive evaluation. It has robustness in illumination transformation, rotation transformation and scale transformation, as well as superior real-time performance. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
