红外, 2009, 30 (9): 35, 网络出版: 2010-05-26
基于旋转不变性的红外序列图像自动匹配
Automatic Matching of Infrared Sequence Images Based on Rotation Invariant
红外序列图像 去噪81FT 旋转不变性 图像匹配 无人机视频 infrared sequence irnage De-noising 81FT rotation invariant irnage rnatching unrnanned Aerial Vehicle (UAV) Video
摘要
红外序列图像的匹配处理是在一系列红外图像内寻求像素点对应关系的过程,它直接涉及并影响到红外图像配准的精度,是视频目标识别、定位、跟踪等处理的关键技术之一。匹配中为了加快搜索速度,采用了金字塔策略。针对红外图像信噪比低的特点,预处理时引入了小波间值算法对红外图像进行滤波处理;针对无人机红外序列图像间旋转角度大的特点,采用81FT 算法对灰度序列图像进行了匹配,并结合相关系数控制的方法,进一步去除错配点。匹配时不受传感平台较大角度旋转的影响,从而取得了良好的效果。
Abstract
The rnatching of infrared sequence irnages is a process to search the corresponding relationship between the pixels in a series of infrared irnages. Since it can directly affect the registration precision of infrared irnages, it is one of the key techniques for the data processing, such as the recognition, location and tracking of an object in video irnages. 1n order to increa启e the search speed, the Pyrarnid 8trategy is used in irnage rnatchi吨. 1n view of the features such as low 8jN ratio of infrared sequence irnages, the wavelet threshold algorithrn is introduced to filter the infrared irnages in preprocessing. For the infrared seque配e irnages of Unrnanned Aerial Vehicles (UAV) , the 81FT algorithrn is used in irnage sequence rnatching and is also cornbined with the correlation coefficient control rnethod so as to get rid of the wrong rnatching points. Because the rnatching is not affected by the rotation of the sensing platforrn at a large angle, good effectiveness is obtained
刘景正, 余旭初, 刘航冶. 基于旋转不变性的红外序列图像自动匹配[J]. 红外, 2009, 30(9): 35. LlU Jing-zheng, YU Xu-chu, LlU Hang-ye. Automatic Matching of Infrared Sequence Images Based on Rotation Invariant[J]. INFRARED, 2009, 30(9): 35.