摘要： 为解决传统尺度不变特征变换(SIFT)算法在光照变化和遮挡的情况下，不能快速准确跟踪目标的问题，提出一种采用粒子滤波和SIFT建立目标模型的方法，利用粒子滤波预测目标在当前帧中可能的位置。计算目标可能存在的区域SIFT特征点，构建特征描述向量，进行目标匹配。根据目标模型和目标候选区域中SIFT特征点的匹配情况，在跟踪过程中更新特征描述向量，实现目标跟踪。实验结果证明，该算法可提高目标检测和跟踪的速度以及准确性。
关键词:
尺度不变特征变换,
粒子滤波,
模型更新,
目标跟踪,
搜索策略,
特征描述向量
Abstract: In the case of illumination varies and shelter, the traditional algorithm can not track the targets fast and precisely. In order to solve the problem, a new method is proposed for target tracking, in which particle filtering is used to establish the target motion model and Scale Invariable Feature Transformation(SIFT) feature is introduced to create the target model. The possible position of the target in the current frame is predicted by particle filtering. The SIFT feature points of the possible area are calculated and the characteristics description vector is built to match the target. According to the matching result of the target model and the SIFT feature points of the target candidate region, the characterization vector is updated in the tracking process. Experimental results show that the proposed algorithm can improve the speed and accuracy of the target detection and tracking significantly.
Key words:
Scale Invariable Feature Transformation(SIFT),
particle filtering,
model update,
object tracking,
search strategy,
characterization vector
中图分类号:
张琪, 张金林, 芮挺, 方虎生. 基于改进SIFT算法的粒子滤波目标跟踪[J]. 计算机工程, 2012, 38(10): 14-17.
ZHANG Qi, ZHANG Jin-Lin, RUI Ting, FANG Hu-Sheng. Particle Filtering Object Tracking Based on Improved SIFT Algorithm[J]. Computer Engineering, 2012, 38(10): 14-17.