摘要
针对背景差分法中的鬼影扰乱运动目标检测与跟踪问题,提出目标相似性度量的鬼影抑制方法。首先分析间隔帧中目标的直方图分布和像素变化率,依此判断目标相似度并检测鬼影;然后提出面向鬼影对象的背景模型更新方法,快速校正鬼影背景,抑制鬼影再出现。实验结果表明,该方法克服了自适应背景方法的检测灵敏度低和运动属性方法消耗高的缺点,能够快速准确地抑制鬼影。
Aiming at the problem that ghost disrupt the accuracy of moving target detection and tracking, this paper proposed ghost suppression method based on object similarity measure. First it analyzed object similarity between interval frame by histo- gram contrast analysis and rate of change pixels to detect ghost, and then adjusted ghost background mode rapidly in the object sample space, restrained ghosts appear again. Experiments show that the proposed method overcome shortcomings of the tradi- tional ghost detection, low sensitivity and high consumption, quickly and accurately detect the ghost targets, and ultimately a- chieve the Dumose of inhibitin~ --host.
出处
《计算机应用研究》
CSCD
北大核心
2014年第3期926-928,932,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(40901197
41161061)
云南省自然科学基金资助项目(2008D032M)
关键词
运动目标检测
鬼影
相似性度量
直方图对比分析
背景模型更新
moving target detection
ghost
similarity measure
histogram contrast analysis
background model update