摘要
利用自然背景与人造目标在分形维数上的差异,提出了一种针对自然背景下的快速红外目标检测技术。通过对图像各个子图的改进不规则分形维数的计算,得到各个子图的指标函数J曲线;经过对J曲线的适当策略分析,可以得到目标最大可能存在的子图区域,或者多目标可能存在的区域。仿真实验证明:算法可以快速检测出红外图像中人造目标可能出现的区域,为随后的识别跟踪等分析降低了计算量。
A fast detection algorithm for artificial IR targets is presented which is based on the difference of fractal dimensions between Simple natural background and man-made targets. By the calculation of the fractal dimension of each sub-image, the J plot of sub-images is made as a goal function. Under the analysis of some preference strategy for J plot, the most possible areas with targets are recommended. The experiment results show that the artificial targets could be quickly detected in maximum possibility, the calculations are reduced in the subsequence recognition analysis.
出处
《航空兵器》
2006年第6期31-33,共3页
Aero Weaponry
关键词
分形维数
自然背景
人造目标
红外图像
目标检测
fractal dimension
natural background
artificial target
infrared image
target detection
