融合速度特征的压缩感知目标跟踪算法

发布时间：2018-06-02 10:15

本文选题：目标跟踪 + 压缩感知　；参考：《南京大学学报(自然科学)》2016年01期

【摘要】：现有基于快速压缩感知的目标跟踪算法采用固定尺寸的搜索框搜索目标,当遇到目标快速移动时容易超出算法的搜索范围,导致跟踪失败.为解决此问题,提出加入目标位移速度特征的快速压缩感知跟踪算法使得搜索目标的范围自适应变化.新方法的思路是首先利用目标在帧间的位移表示出目标的位移速度,然后将当前帧内的目标位移速度与前几帧的平均速度相比较,再根据目标位移速度变化自适应改变搜索范围,即当目标运动速度保持稳定则保持搜索框尺寸,目标运动速度加快则增大搜索框尺寸,目标运动速度变慢则缩小搜索框尺寸,以适应目标移动速度的变化.在目标快速移动的视频集上的实验结果显示,新方法自适应地改变搜索范围,一直都能跟踪到目标,特别是当现有的压缩感知跟踪算法丢失目标时,新方法仍能比较好地跟踪到目标.
[Abstract]:The existing target tracking algorithm based on fast compression perception uses a fixed-size search box to search the target. When the target moves quickly, it is easy to exceed the search range of the algorithm, resulting in the tracking failure. In order to solve this problem, a fast compression perceptual tracking algorithm with target displacement velocity feature is proposed to make the range of search targets adaptive. The idea of the new method is to first express the displacement velocity of the target by using the displacement of the target between frames, and then compare the displacement velocity of the target in the current frame with the average velocity of the previous frames. Then adaptively changing the search range according to the change of the target displacement velocity, that is, keeping the size of the search frame when the moving speed of the target remains stable, increasing the size of the search frame if the moving speed of the target increases, and reducing the size of the search frame if the moving speed of the target becomes slower. To adapt to changes in the moving speed of the target. Experimental results on a fast moving video set show that the new method adaptively changes the search range and can always track the target, especially when the existing compressed perceptual tracking algorithm loses the target. The new method can track the target well.
【作者单位】：福建师范大学数学与计算机科学学院福建省网络安全与密码技术重点实验室;
【基金】：国家自然科学基金(61175123) 福建师范大学“网络与信息安全关键理论和技术”校创新团队项目(IRTL1207)
【分类号】：TP391.41

【参考文献】