一种VideoSAR动目标阴影检测方法
发布时间：2019-03-15 09:51
【摘要】：在高帧率的视频合成孔径雷达(VideoSAR)成像模式获得的图像序列中,多普勒频移使运动目标在实际位置留下阴影,且相邻帧图像具有很强相关性。该文针对上述现象提出一种VideoSAR图像中动目标阴影检测的方法。首先,对每帧图像通过结合尺度不变特征变换(SIFT)和随机抽样一致性(RANSAC)算法实现配准并进行背景补偿,再采用CattePM模型抑制相干斑噪声。然后通过Tsallis灰度熵的最大化阈值分割方法自动分离目标和背景,获得二值图像。最后,对相邻多帧图像背景建模并差分,再结合三帧间差分法提取动目标阴影,结果标记至原帧图像相应位置。基于美国Sandia实验室公布的VideoSAR成像片段,实现了多个移动车辆的检测,验证了所提算法的有效性。
[Abstract]:In the high frame rate video synthetic aperture radar (VideoSAR) imaging image sequence, Doppler frequency shift causes the moving target to leave shadow in the real position, and the adjacent frame image has strong correlation. In this paper, a method for shadow detection of moving targets in VideoSAR images is proposed. Firstly, each frame is registered and compensated by combining scale invariant feature transform (SIFT) and random sampling consistent (RANSAC) (RANSAC) algorithm, and then the CattePM model is used to suppress speckle noise. Then the object and background are automatically separated by the maximum threshold segmentation method of Tsallis gray entropy, and the binary image is obtained. Finally, the background of the adjacent multi-frame image is modeled and differentiated, and then the moving object shadow is extracted by using the three-frame difference method, and the result is marked to the corresponding position of the original frame image. Based on the VideoSAR image fragment published by American Sandia Laboratory, the detection of several mobile vehicles is realized, and the effectiveness of the proposed algorithm is verified.
【作者单位】： 南京航空航天大学电子信息工程学院;南京工程学院计算机工程学院;
【基金】：国家自然科学基金(61671240) 江苏省自然科学基金青年基金(BK20150730) 中央高校基本科研业务费(NZ2016105) 南京航空航天大学研究生创新基地(实验室)开放基金资助项目(kfjj20170401)~~
【分类号】：TN957.52
本文编号：2440522
[Abstract]:In the high frame rate video synthetic aperture radar (VideoSAR) imaging image sequence, Doppler frequency shift causes the moving target to leave shadow in the real position, and the adjacent frame image has strong correlation. In this paper, a method for shadow detection of moving targets in VideoSAR images is proposed. Firstly, each frame is registered and compensated by combining scale invariant feature transform (SIFT) and random sampling consistent (RANSAC) (RANSAC) algorithm, and then the CattePM model is used to suppress speckle noise. Then the object and background are automatically separated by the maximum threshold segmentation method of Tsallis gray entropy, and the binary image is obtained. Finally, the background of the adjacent multi-frame image is modeled and differentiated, and then the moving object shadow is extracted by using the three-frame difference method, and the result is marked to the corresponding position of the original frame image. Based on the VideoSAR image fragment published by American Sandia Laboratory, the detection of several mobile vehicles is realized, and the effectiveness of the proposed algorithm is verified.
【作者单位】： 南京航空航天大学电子信息工程学院;南京工程学院计算机工程学院;
【基金】：国家自然科学基金(61671240) 江苏省自然科学基金青年基金(BK20150730) 中央高校基本科研业务费(NZ2016105) 南京航空航天大学研究生创新基地(实验室)开放基金资助项目(kfjj20170401)~~
【分类号】：TN957.52
【相似文献】
相关博士学位论文 前1条
1 代江艳;视频中运动目标阴影检测研究[D];东北师范大学;2014年
相关硕士学位论文 前1条
1 刘丽红;视频监控中运动目标的检测、分类识别与跟踪[D];湖南工业大学;2016年
，本文编号：2440522
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2440522.html
