基于过渡区提取的红外运动目标检测  

Infrared moving target detection based on transition region extraction

作　　者：许文武[1] 段先华[1] 

机构地区：[1]江苏科技大学计算机科学与工程学院,江苏镇江212003

出　　处：《激光与红外》2010年第7期775-778,共4页Laser & Infrared

摘　　要：针对复杂动态场景下的红外目标检测问题,提出了一种基于交叉熵的过渡区提取的红外运动目标检测方法。该方法首先使用帧差法和背景差分法相融合的检测方法,对红外图像进行差分处理,然后采用基于交叉熵的过渡区分割算法二值化图像,最后进行形态学滤波,从而检测出完整的红外目标。实验结果表明,该方法目标检测效果比较好,能满足红外运动目标检测的需要。Regarding the infrared target detection under the complex dynamic scene,a method is proposed for infrared moving target detection based on cross-entropy transition region extraction.Firstly,infrared image is processed with method of frame difference and background difference.Then binary image is obtained by segmentation algorithm based on cross-entropy transition region extraction.Finally,the complete infrared target is detected by the morphology filter.The experimental result indicates that this method is quite fit for infrared target detection and satisfies infrared moving target detection.

关 键 词：红外目标检测 交叉熵 过渡区 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]