论文部分内容阅读
针对数字地图中的线状特征,如道路、河流的提取问题,提出一种基于序列蒙特卡洛的特征提取算法。算法利用粒子滤波方法与贝叶斯模型的建模思路,描述数字地图道路的状态及其观测估计方法;在此基础上,讨论跟踪过程中的模拟退火处理以及粒子聚类分析;最后,将跟踪模型应用在数字地图的特征提取中。实验表明,与传统的数字栅格地图道路提取算法以及基于其它粒子滤波的算法相比较,本文提出的基于跟踪模型的算法具有较强的鲁棒性以及精度,同时能够很好地处理扫描栅格地图中缺失的道路信息提取。
Aiming at the linear features in digital maps, such as the extraction of roads and rivers, a feature extraction algorithm based on sequence Monte Carlo is proposed. The algorithm uses the particle filter and the Bayesian model to describe the state of the digital road and its observation and estimation method. On this basis, the simulated annealing process and the particle clustering analysis are discussed. Finally, the tracking The model is used in the feature extraction of digital map. Experiments show that compared with the traditional digital raster map road extraction algorithm and other particle filter-based algorithms, the proposed tracking model-based algorithm has strong robustness and accuracy, and can well handle the scanning raster Missing road map information extraction.