一种复杂环境下运动目标检测的背景构建方法

¹烟台职业学院图像处理研究所,烟台市滨海中路2018号264670;²海军航空工程学院电子信息工程系,烟台市二马路188号,264001;³清华大学电子工程系,北京市清华园100084

基金项目: 国家自然科学基金资助项目(60541001)；全国优秀博士学位论文作者专项基金资助项目(200443)；“泰山学者”建设工程专项经费资助项目；山东省高校优秀青年教师国内访问学者基金资助项目

详细信息

Background Building Method of Moving Objects Detection in Complex Scenes

¹Laboratory of Image Processing,Yantai Vocational College,2018 Middle Binhai Road,Yantai 264670,China;(2 Department of Electronic Engineering,Tsinghua University,Tsinghuayuan,Beijing 100084,China;(3 Department of Electronic Information,Naval Aeronautical Engineering Institute,188 Erma Road,Yantai 264001,China

Funds: 国家自然科学基金资助项目(60541001)；全国优秀博士学位论文作者专项基金资助项目(200443)；“泰山学者”建设工程专项经费资助项目；山东省高校优秀青年教师国内访问学者基金资助项目

摘要: 针对复杂环境下背景构建困难等问题,提出了一种彩色图像依照目标运动状态实时进行更新的背景构建方法。该方法通过定义彩色图像距离,根据当前图像与背景图像之间的距离变化定义运动状态矩阵,从而利用状态矩阵结合当前帧图像的彩色特征更新背景。该方法无需假设背景模型,能够适应各类复杂的突变,通过获取包含多类目标的实测彩色图像实验验证了算法的有效性和准确性。
- 运动目标检测 /
- 背景差 /
- 运动状态矩阵 /
- 更新
Abstract: We present a background image building method to deal with the problem of color images motion object detection in real-time.The distance between color images is defined to compare the frame image and the background image according to the frame sequences.The motion state matrix is extracted to update background image in pixels level.The method need not background models and fit variety changes.Actual color images were used to test the method.The experimental results show that our method is effective and accurate.
- motion object detection /
- background subtraction /
- motion state matrix /
- update