| 基于声呐图像序列SIFT特征的多目标跟踪方法研究 |
| |
| 引用本文： | 张丽红,周天,徐超,韩婷婷.基于声呐图像序列SIFT特征的多目标跟踪方法研究[J].声学技术,2019,38(5):514-519. |
| |
| 作者姓名： | 张丽红 周天 徐超 韩婷婷 |
| |
| 作者单位： | 哈尔滨工程大学水声技术重点实验室, 黑龙江哈尔滨 150001;海洋信息获取与安全工信部重点实验室(哈尔滨工程大学), 工业和信息化部, 黑龙江哈尔滨 150001;哈尔滨工程大学水声工程学院, 黑龙江哈尔滨 150001,哈尔滨工程大学水声技术重点实验室, 黑龙江哈尔滨 150001;海洋信息获取与安全工信部重点实验室(哈尔滨工程大学), 工业和信息化部, 黑龙江哈尔滨 150001;哈尔滨工程大学水声工程学院, 黑龙江哈尔滨 150001,哈尔滨工程大学水声技术重点实验室, 黑龙江哈尔滨 150001;海洋信息获取与安全工信部重点实验室(哈尔滨工程大学), 工业和信息化部, 黑龙江哈尔滨 150001;哈尔滨工程大学水声工程学院, 黑龙江哈尔滨 150001,哈尔滨工程大学水声技术重点实验室, 黑龙江哈尔滨 150001;海洋信息获取与安全工信部重点实验室(哈尔滨工程大学), 工业和信息化部, 黑龙江哈尔滨 150001;哈尔滨工程大学水声工程学院, 黑龙江哈尔滨 150001 |
| |
| 基金项目： | 国家自然科学基金（U1709203、41376103）资助项目。 |
| |
| 摘 要： | 针对高频成像声呐多目标跟踪中声成像的不稳定问题，在基于检测前跟踪的声呐图像序列尺度不变特征变换（Scale Invariant Feature Transform，SIFT）的基础上，提出了一种结合对数变换的声呐图像序列SIFT特征跟踪方法。该方法利用对数变换进行图像预处理，进而利用声呐图像序列SIFT特征进行多目标跟踪。水池试验数据的处理结果表明：与中值滤波和动态亮度分配的方法相比，结合对数变换的声呐图像序列SIFT特征跟踪方法具有更好的多目标跟踪效果。
|
| 关 键 词： | 声呐图像序列 检测前跟踪 尺度不变特征变换 图像预处理 对数变换 多目标跟踪 |
| 收稿时间： | 2018/5/9 0:00:00 |
| 修稿时间： | 2018/6/28 0:00:00 |
Research on multi-target tracking method with SIFT features of sonar image sequences |
| |
| ZHANG Li-hong,ZHOU Tian,XU Chao and HAN Ting-ting.Research on multi-target tracking method with SIFT features of sonar image sequences[J].Technical Acoustics,2019,38(5):514-519. |
| |
| Authors: | ZHANG Li-hong ZHOU Tian XU Chao and HAN Ting-ting |
| |
| Affiliation: | Acoustic Science and Technology Laboratory, Harbin Engineering University, Harbin 150001, Heilongjiang, China;Key Laboratory of Marine Information Acquisition and Security(Harbin Engineering University), Ministry of Industry and Information Technology, Harbin 150001, Heilongjiang, China;College of Underwater Acoustic Engineering, Harbin Engineering University, Harbin 150001, Heilongjiang, China,Acoustic Science and Technology Laboratory, Harbin Engineering University, Harbin 150001, Heilongjiang, China;Key Laboratory of Marine Information Acquisition and Security(Harbin Engineering University), Ministry of Industry and Information Technology, Harbin 150001, Heilongjiang, China;College of Underwater Acoustic Engineering, Harbin Engineering University, Harbin 150001, Heilongjiang, China,Acoustic Science and Technology Laboratory, Harbin Engineering University, Harbin 150001, Heilongjiang, China;Key Laboratory of Marine Information Acquisition and Security(Harbin Engineering University), Ministry of Industry and Information Technology, Harbin 150001, Heilongjiang, China;College of Underwater Acoustic Engineering, Harbin Engineering University, Harbin 150001, Heilongjiang, China and Acoustic Science and Technology Laboratory, Harbin Engineering University, Harbin 150001, Heilongjiang, China;Key Laboratory of Marine Information Acquisition and Security(Harbin Engineering University), Ministry of Industry and Information Technology, Harbin 150001, Heilongjiang, China;College of Underwater Acoustic Engineering, Harbin Engineering University, Harbin 150001, Heilongjiang, China |
| |
| Abstract: | The instability of acoustic imaging is still a challenge in multi-target tracking of high-frequency imaging sonar. Based on the SIFT (scale-invariant feature transform) feature tracking method using TBD (track-before-detect) technique, a SIFT feature tracking method with logarithm transformation for sonar image sequences is proposed. This method uses logarithmic transformation to preprocess images, and then uses the SIFT features of sonar image sequence to perform multi-target tracking. Logarithmic transformation can compress high gray-scale pixel values to improve the detection ability of weak targets and the stability of sonar imaging. The results of pool experiment show that the SIFT feature tracking method with logarithm transformation performs better in multi-target tracking of sonar image sequences, compared with the median filtering and dynamic brightness allocation method. |
| |
| Keywords: | sonar image sequences track-before-detect scale-invariant feature transform image preprocessing logarithmic transformation multi-target tracking |
|
| 点击此处可从《声学技术》浏览原始摘要信息 |
|
点击此处可从《声学技术》下载全文 |
|
