论文部分内容阅读
针对移动机器人在目标识别过程中对视觉图像特征点提取慢,匹配不准确等特点,提出了一种基于SIFT算法的改进目标识别算法。通过采用组合匹配策略,将特征关键点间的距离和内积同时进行考察,根据其自身值的大小,决定对匹配相似度的贡献。组合策略的引入有效地解决了机器人在目标识别中对相同特征图像不能匹配和不同特征图像能够匹配的问题。为克服目标匹配时实性差的弱点,以关键点为根据构建K维树结构,采用最近邻点搜索,快速找出正确匹配的特征点。为实现移动机器人目标识别过程中的自主性,在特征点匹配过程中引入自适应阈值进行判断。实验表明,该方法对移动机器人目标识别准确率有较大提升,能够满足移动机器人在目标识别和跟踪过程中对视频图像处理的实时性和准确性的要求。
Aiming at the characteristics of moving robot such as slow extraction of feature points of visual images and inaccurate matching during target recognition, an improved target recognition algorithm based on SIFT algorithm is proposed. Through the use of combinatorial matching strategy, the distance and inner product between feature keys are investigated simultaneously, and the contribution to matching similarity is decided according to the value of its own value. The introduction of the combined strategy effectively solves the problem that the robot can not match the same feature image and different feature images in the target recognition. In order to overcome the weakness of poor solidarity when the target is matched, the K-dimensional tree structure is constructed based on the key points, and nearest neighbor search is used to quickly find the correct matching feature points. In order to realize the autonomy in the process of target recognition of mobile robot, adaptive threshold is introduced in feature point matching. Experiments show that this method can greatly improve the accuracy of target recognition for mobile robots and meet the requirements of real-time and accuracy of video image processing in the process of target recognition and tracking.