期刊文献+

概率加权质心跟踪算法研究 被引量:7

Probability weighted centroid tracking algorithm
原文传递
导出
摘要 针对目前图像跟踪器跟踪不稳定、跟踪精度不高及不能满足实时性要求等问题,提出了一种概率加权的质心跟踪算法。该算法首先对波门内的像素进行阈值分割,摒弃灰度低于阈值的背景像素,保留目标像素的灰度值,然后计算波门内目标区域的质心。实验结果表明:基于概率加权的质心跟踪方法能够有效降低复杂背景和噪声干扰,增强跟踪系统的抗干扰能力,减少传统跟踪系统中使用大量灰度梯度值带来的巨大计算量,从而提高跟踪器的精度和稳定性。创新点在于通过引入概率加权的方法,在计算初始时刻的目标质心时使用贝叶斯概率作为权重,而没有设置离散的阈值来分辨目标,减少了传统跟踪系统中使用大量灰度梯度值产生的计算复杂度。 The probability weighted centroid tracking(PWC) algorithm was proposed to solve the problems of the image tracker,such as the low tracking precision,instability and could not meet the requirement of real time. Firstly,the threshold segmentation of the pixels in wave-gate was processed by the algorithm.These pixels whose gray levels were lower than the counterparts of background were discarded . The gray values of targets pixels were retained. And then the centroid coordinates of wave-gate in the targets regions were obtained. The experimental results demonstrate that PWC can effectively decrease the background clutter and noise countermeasure as well as reduce huge calculation and computation complexity due to computing multiple graduations of gray levels in traditional tracking methods of tracker. Furthermore,it can also increase the ability of anti-countermeasure and greatly improve the tracking precision and stability of the tracker. The Bayesian probability was introduced and regarded as the weight when calculating the initiative time of the target centroid.
出处 《红外与激光工程》 EI CSCD 北大核心 2008年第4期621-624,共4页 Infrared and Laser Engineering
基金 武器装备预研基金项目(41302020103)
关键词 图像处理 概率加权 质心跟踪 目标检测 Image processing,Probability weighted,Centroid tracking,Target detecting
  • 相关文献

参考文献3

二级参考文献13

  • 1崔玉平,刘永才.新的红外图像小目标检测方法(英文)[J].红外与激光工程,2005,34(5):587-591. 被引量:7
  • 2[1]T.I. Hsu, J. L. Kuo and R. Wilson. A multiresolution texture gradient method for unsupervised segmentation,Pattern Recognition 2000, 33(11):1819~1834
  • 3[2]N.P. Pal and S. K. Pal. A review on image segmentation techniques, Pattern Recognition, 1993, 26(9): 1277~1294
  • 4[3]R. Bergevin and M. Mokhtari. Multiscale contour segmentation and approximation: an algorithm based on geometry of regular inscribed polygons, CGIU, 1998,71(1):55~73
  • 5[4]A.S.E. Koster, K. L. Vincken, C. N. de Graaf, O. C.Zander and M. A. Viergever. Heuristic linking models in multiscale image segmentation, CGIU, 1997,65(3):361~381
  • 6[5]C.L. Wilson, C. I. Watson and E. G. Pack. Effect of resolution and image quality on combined optical and neural network fingerprint matching, Pattern Recognition,2000, 33(2):317~332
  • 7[6]V. Beau and M. Singer. Reduced resolution and scale space for dominant feature detection in contours, Pattern Recognition, 2001, 34(2):287~298
  • 8刘国衡,关玉才.图象序列的目标检测[J].计算机工程与应用,1997,33(8):26-29. 被引量:5
  • 9邓英士,王敬儒.一种快速图象处理相关匹配方法[J].光电工程,1997,24(4):6-10. 被引量:5
  • 10杨树谦.精确制导武器和技术[J].红外与激光工程,1999,28(6):1-3. 被引量:19

共引文献30

同被引文献71

引证文献7

二级引证文献29

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部