基于小波子带模型匹配的同水域下目标探测

发布时间：2018-06-17 07:19

本文选题：声呐 + 水下目标探测　；参考：《电子测量与仪器学报》2017年11期

【摘要】：针对水下声呐目标搜寻过程中环境复杂、成像分别率低、人工识别目标困难等实际问题,提出基于双树双密度复小波子带模型匹配的方法自动判别水下目标有无。首先利用双树双密度复小波对所传回待测声图进行分解得到不同方向子带,接着采用三参数广义Γ函数对不同方向子带系数分布拟合作为特征项,最后利用含有目标声图与背景声图在分布特征上存在差异性,根据计算两者间KL距离进行相似度匹配并通过设定最优门限对声图目标有无自动判别。实验证明,利用所提出方法对含有多目标声呐图像判别,其检测品质因子达97.2%,并在不同噪声水平下其检测品质因子对比其他方式平均高出10%,具有较高的检出率和一定鲁棒性。
[Abstract]:In order to solve the practical problems in underwater sonar target search, such as complex environment, low imaging rate and difficulty in identifying targets manually, a method based on double tree and dual density complex wavelet subband model matching is proposed to automatically identify underwater targets. In this paper, we first decompose the acoustic images to different directions by using double tree and double density complex wavelets, and then use three parameter generalized 螕 function to fit the distribution of coefficients in different directions. Finally, based on the difference of the distribution characteristics between the sonogram with the target and the background, the KL distance between the two is calculated to match the similarity, and the optimal threshold is set to determine whether there is an automatic recognition of the target in the acoustic image. The experimental results show that the detection quality factor of sonar images with multiple targets is 97.2 by using the proposed method. The detection quality factor of sonar images with different noise levels is 10 times higher than that of other methods, so it has a high detection rate and a certain robustness.
【作者单位】：南京信息工程大学江苏省气象探测与信息处理重点实验室;南京信息工程大学江苏省气象传感网络技术工程中心;南京邮电大学电子科学与工程学院;
【基金】：江苏省第11批六大高峰人才项目(2014-XXRJ-006) 江苏省重点研发计划社会发展项目(BE201569) 国家自然科学基金(41075115) 江苏高校优势学科Ⅱ期建设工程项目资助
【分类号】：TB56;TP391.41

【相似文献】