摘要
关于视频图像多目标检测优化问题,针对雾天视频图像目标特征和背景不断地变化,雾天图像的退化程度跟场景深度成非线性关系的问题,提出用于准确完成雾天运动目标检测的实时背景建模框架的图像清晰化方法,并创建一种新的具有增量集成学习能力的目标跟踪计算模型和相应的方法,以适应跟踪过程中目标特征和背景的不断变化,构建一种具有增量学习能力的稳健的目标跟踪算法,改进方法有效地解决了雾天条件下运动目标跟踪的稳健性。
About multi-objective optimization problem detection for video image.Aim at foggy day video image target feature and the background constantly changing.The problem of the fog image degradation into a non-linear relationship with the scene depth question.Proposed for accurately complete real-time background modeling framework for image clarity methods of fog moving target detection.And create a new target tracking mathematical model and corresponding method with incremental learning ability.To accommodate tracking process of tget feature and the background changing.Building a kind of robust object tracking algorithm with incremental learning capability.Improved method effective solution to the moving target tracking robustness in foggy conditions.
出处
《计算机仿真》
CSCD
北大核心
2013年第8期402-407,430,共7页
Computer Simulation
基金
国家科技支撑计划项目2012 BAI34 B00(Grant No.2012BAI34B00)
安徽高校省级自然科学研究一般项目(KJ2012Z325)
关键词
图像增强
视觉计算
增量集成学习
目标跟踪
Image enhancement
Visual computing
Learning of adaboost incremental integrated
Target tracking
