基于背景重构的运动对象越界侦测方法  

Motion Object Cross-border Detection Method Based on Background Reconstruction

作　　者：洪虹[1] 李文耀[1] 

机构地区：[1]武汉邮电科学研究院,湖北武汉430074

出　　处：《电视技术》2012年第7期123-126,共4页Video Engineering

摘　　要：提出了一种基于背景重构的运动对象越界侦测方法,该算法利用当前视频图像和背景视频图像,通过差分法获取运动对象模型和背景差值,再利用运动对象模型的连续性绘制运动对象轨迹,对轨迹和警戒线交叉方程进行越界侦测。同时,利用当前视频图像融合背景图像生成新的背景图像,能有效保证识别算法的自适应性,提高侦测结果的准确性。A method of motion object detection based on the background of the reconstruction is proposed. The algorithm gets motion object models and background difference through the difference method by current video images and video image background, and draws objects track by moving object model, for cross-border detection by track and cross equation. At the same time, it can effectively ensure the adaptability of recognition algorithm, and enhance the accuracy of the detection results by new background image which is generated by current video image and background image.

关 键 词：智能视频 运动对象检测 越界侦测 

分 类 号：TN919.8[电子电信—通信与信息系统]