计算机科学 ›› 2012, Vol. 39 ›› Issue (4): 265-268.

• 图形图像 • 上一篇    下一篇

多线程下的视频微动目标检测与提取算法研究

瞿 中,马庆伟,张庆庆   

  1. (重庆邮电大学计算机科学与技术学院 重庆400065);(重庆邮电大学移通学院 重庆401520)
  • 出版日期:2018-11-16 发布日期:2018-11-16

Research on the Algorithm of Micro-movement Target Tracking and Extraction in Video by Multi-thread Technology

  • Online:2018-11-16 Published:2018-11-16

摘要: 在多线程并发情况下提取实时视频图像中的微动人物目标,基于多线程的思想,在多路并发请求的情况下,运用肤色检测和聚类思想对图像中的人物进行提取分割,并对处理过程中的多路访问资源的问题进行访问冲突避免。首先采用边缘检测的思想对运动目标边缘进行范围定位,然后采用聚类算法和肤色检测等方法进行目标模板的提取和完整性补充,根据得到的目标模板结合原始图片对应像素点坐标,将原始像素点颜色着色到新的背景模型的对应位置。仿真结果表明,在保证处理质量的情况下,提出的基于多线程思想的算法能满足实际工程应用实时处理的要求。

关键词: 多线程,微动目标,聚类算法,肤色检测,目标模板

Abstract: We extracted micro-movement target of real-time video images under the multi-thread conditions. Based on the multi thread ideas, we used skin colour detection and clustering methods to process and solve the access violation condition. hhe ideology of edge detection was used locate the range of the moving object, and then clustering algorithm and skin color detection and some other methods were used extract the object template and complement the integrity of the object template, and according to the object template and corresponding pixel coordinates of the original image, the original pixel color onto the corresponding position of new background model. The simulation results show that the proposed method ensure the quality requirements of real-time processing and has a certain robustness, so this method satisfies the needs of the project.

Key words: Multi-thread, Micro-movement target, Clustering algorithm, Skin colour detection, Object template

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!