摘要
为了从红外热图像中识别和跟踪目标,提出了一种采用数学形态学和小波变换相结合的红外图像处理新方法。该方法首先利用形态开、闭滤波器组对原始图像进行平滑处理;然后采用一种改进的数学形态学的分水岭算法对红外图像进行分割,同时利用考虑了图像纹理信息的小波阈值对分割的图像进行滤波处理;最后针对过分割问题提出了一种利用区域灰度中值对分割区域进行融合处理的算法。实验结果及算法性能评估结果表明,该方法能够较好地解决红外目标的识别问题,具有较好的实用性。
In the purpose of recognizing and tracking targets from infrared thermal images, an infrared image processing algorithm based on mathematical morphology and wavelet transform was presented. A original image was smoothed with the open and close filters of mathematical morphology, then the infrared image was segmented with improved watershed, simultaneously, the segmented image was processed with the wavelet threshold with picture texture information. For the problem of over-seg＇ mentation regions, an effective region merging method was settled to fuse the segmentation region using the region gray median. The experiment results and algorithm performance evaluation show that the method performs well in infrared target recognition and has a good practicability.
出处
《光学精密工程》
EI
CAS
CSCD
北大核心
2007年第1期138-144,共7页
Optics and Precision Engineering
基金
总装预研项目(No.6140403)
关键词
分水岭算法
小波闽值滤波
红外目标检测
红外图像处理
watershed algorithm
wavelet threshold filter
infrared target detection
infrared imageprocessing
作者简介
同武勤（1980-），男，陕西韩城人．博士研究生，研究方向为毫米波电子对抗，红外／雷达目标跟踪、识别及图像处理；E-mail：tongwuqin@sogou．com
凌永顺（1937-），男，安徽定远人，教授，中国工程院院士，研究方向为电子工程。
