Self-Constructed Sliding Kernel Correlation Algorithm for Underwater Weak Target Detection via Active Sonar Under Strong Clutter Interference

Xingyue Zhou, Wutao Yin, Kunde Yang

Research output: Contribution to journalArticlepeer-review

Abstract

With the expanding application of active sonar, detecting underwater moving targets in complex shallow water environments with low false alarm rates remains a significant challenge due to strong reverberation and obscuring clutter. Although low-rank sparse decomposition (LRSD) can mitigate large-area near-end reverberation, it is less effective in addressing weak fluctuation clutter. To address this issue, a self-constructed sliding kernel correlation (SSKC) algorithm is proposed to detect target under clutter sources such as reefs, nets, and wrecks. First, the original echo is segmented through sliding window, meantime, the sliding kernel is constructed via kernel density estimation (KDE). Simultaneously, another kernel is designed based on the linear frequency modulation transmitted signal, and the filtered matrices processed by the two kernels are beamformed, respectively. Then, the kernel correlation matrix is calculated and applied to LRSD operation is applied to further restrain reverberation. Finally, the matrix similarity between the sparse matrix and the kernel correlation matrix is extracted to detect weak moving objects. Experimental results in shallow water demonstrate that true-positive rate (TPR) of the proposed method can reach to 0.91 under the condition of false-positive rate (FPR) of 0.1.

Original languageEnglish
Pages (from-to)1-10
Number of pages10
JournalIEEE Transactions on Instrumentation and Measurement
Volume74
DOIs
StatePublished - 2025

Keywords

  • Active sonar measurement
  • clutter suppression
  • self-constructed kernel
  • underwater weak target detection

Fingerprint

Dive into the research topics of 'Self-Constructed Sliding Kernel Correlation Algorithm for Underwater Weak Target Detection via Active Sonar Under Strong Clutter Interference'. Together they form a unique fingerprint.

Cite this