摘要
针对遥感影像智能判读领域中如何提高目标检测效率的问题,将视觉显著模型应用到基于特征统计学习的目标检测中,首先通过计算视觉显著图,提取飞机目标潜在区域,然后设计分类器完成目标检测任务。实验结果表明,该方法与传统的遍历搜索方法相比,大幅减少了搜索空间,降低了误检率,在检测率不变的情况下提高了50%的检测效率。
According to the problem how to improve the object detection efficiency in the domain of intelligence discrimination for remote sensing image, this paper adopts visual salient model to object detection based on characteristic statistical learning, firstly extracts the latent area of aircraft object by calculating visual salient image;then designs classifier to achieve the object detection mission. Experiment results show that this method greatly reduces the searching space, decreases the miss- detection probability, and improves detection efficiency 50% in the condition of invariable detection probability compared with the method based on traditional ergodic searching.
出处
《舰船电子对抗》
2013年第5期46-50,87,共6页
Shipboard Electronic Countermeasure
关键词
视觉显著模型
搜索策略
级联分类器
目标检测
visual salient model
searching strategy
cascade classifier
target detection
