论文部分内容阅读
在受生物导航方式启发的局部视觉归航算法中,ALV(average landmark vector)算法因其模型简单、归航性能较好以及所需存储空间小等优点受到了广泛的重视.在非结构化环境中,ALV算法常常需要使用图像的局部特征作为自然路标点,在这种情况下,路标的对应性问题难以保证,同时在保证归航性能的前提下如何合理地精简路标数量也尚无有效的解决方法.针对上述问题,对基于双曲面镜的折反射全景图像进行了研究,提出了horizon环域的概念.在环域内提取SIFT(scale invariant feature transform)特征作为自然路标点并结合ALV模型提出了一种改进的基于自然路标的ALV算法.改进算法有效地缩小了路标点的提取区域,较好地保证了路标点的对应性并精简了路标点的数量.多个实际场景的实验表明,这种算法有效提高了归航精度.
In the local visual navigation algorithm, which is inspired by biological navigation, the algorithm of average landmark vector (ALV) is widely valued due to its simple model, good navigation performance and small storage space required.In the unstructured environment , The ALV algorithm often needs to use the local features of the image as the natural waypoint. In this case, the problem of the correspondingness of the roadmap is difficult to guarantee. At the same time, how to reasonably reduce the number of road signs under the premise of homing performance is not effective In order to solve the above problems, the hyperbolic mirror-based panoramic image is studied, and the concept of horizon horizon is proposed. The scale invariant feature transform (SIFT) is extracted as natural landmark and combined with ALV model An improved ALV algorithm based on natural landmark is proposed.The improved algorithm effectively narrows the extraction area of the marker points and ensures the correspondence of the marker points and simplifies the number of the marker points.Experimental results show that, This algorithm effectively improves the homing accuracy.