摘要
为了实现在移动设备中实时的进行目标追踪,设计了一种基于Camshift目标追踪算法在Android平台下的目标追踪系统。通过对Camshift算法进行移植,使用Android NDK把移植后的Camshift算法编译成.so文件,然后利用JNI技术对加载后的.so文件在目标追踪模块进行本地调用。为了使系统更加智能,在追踪模块中引入HOG描述子用于检测第一帧出现的行人。当检测目标是行人时可以自动的定位目标进行追踪。目标是其他物体时就需要进行选中,以获取目标的特征进行跟踪。最终经过对行人和硬币等目标进行不同场景下的测试,对实现结果展开对比,验证系统可以在第一帧视频中检测到行人然后自动的进行追踪,并对手动选择的目标在特定条件下有良好的追踪效果。由此系统实现了在移动设备中的实时目标追踪。
In order to achieve target tracking in mobile devices in real time, a target tracking system based on tSamshift target tracking algorithm is designed. Transplant the Camshift algorithm, use the Android NDK to compile the migrated Camshift algorithm into the. so file and and use JN1 technology to make local calls to the loaded. In the target tracking module, the tracking target is tracked and processed. In order to make the system more intelligent, a HOG descriptor is introduced in the tracking module to detect pedestrians in the first frame. When the detection target is pedestrian, the target can be tracked automatically. When the target is other objects, it will track the characteristics of the target. Final- ly, pedestrian and coin targets are tested under different scenarios. The verification system can detect pedestrians in the first frame and track them automatically by comparing the result of the implementation, and has a good tracking effect on manually selected targets. So. The system realizes real-time target tracking in mobile devices.
作者
卫建华
王亚峰
廉继红
景皛皛
Wei Jianhua;Wang Yafeng;Lian Jihong;Jing Xiaoxiao(School of Electronics and Information,Xiran Polytechnic University,Xiran 710048,China)
出处
《国外电子测量技术》
2018年第10期135-138,共4页
Foreign Electronic Measurement Technology
