摘要
基于复杂天空背景下红外小目标的特性分析,提出了一种利用分类背景预测与图像分块技术进行红外小目标检测的有效算法。该算法以背景预测理论为基础,通过边缘检测技术、最大均值和局部最相似分类背景预测技术获得较为准确的图像背景,进而采用统计分割算法从残差图像中提取出较为精确的目标位置。其中,通过图像分块处理,提高了算法计算效率。最后,选取了三组具有代表性的红外序列对算法的性能进行了检验。实验结果表明,所提出预测算法在检测准确性、鲁棒性以及计算效率上都具有明显的优越性。
The paper presents an efficient algorithm to detect small infrared target based on the principle of background prediction and image blocking according to the feature of small IR target in a complex background.Using the edge detection,maximum mean algorithm and partial similar methods,this approach can obtains a more accurate background.With the adoption of the statistical segmentation method,the proposed method can obtain a more accurate position of target from the residual images.In the process,the image block technique is used to improve the efficiency of the algorithm.Finally,three typical series of IR video were selected to verify the performance of the algorithm.The computer simulations show that the presented algorithm has more advantages in the detection accuracy,robustness,and calculation efficiency.
出处
《激光与红外》
CAS
CSCD
北大核心
2012年第5期572-578,共7页
Laser & Infrared
基金
国家自然科学专项基金项目/科学仪器基础研究专款(No.20927001)
陕西省自然科学基金项目(No.2006F47)
陕西师范大学优秀科技预研基金项目(No.20092006)资助
关键词
红外小目标
边缘检测
图像分块
背景预测
small IR target
edge detection
image block
background prediction
作者简介
郭张婷（1986-）,女,硕士研究生,主要研究方向为目标跟踪及图像处理.E-mail：angelting002@yahoo.com.cn
