Scholarlink

<p>Moldy-core pear has no obvious symptoms in the early stages. The accuracy and reliability of single-point excitation and single-point sensing detection using acoustic vibration methods are low. To further enhance the detection accuracy of moldy-core disease in pears, we propose a new strategy using single-point excitation and dual-point sensing. This approach addresses challenges from internal tissue heterogeneity and anisotropic wave propagation in pears. We collected time-domain signals from the equatorial and calyx regions using self-built acoustic vibration devices. By integrating signals from different channels to enhance the accuracy and reliability of early moldy-core disease detection, the vibration response reflects the dynamic mechanical behavior of pear tissue, and internal moldy-core damage alters stiffness and damping characteristics, leading to changes in resonance behavior and time–frequency energy distribution. We transformed these signals into time–frequency images (TFIs) using short-time Fourier transform (STFT), continuous wavelet transform (CWT), and Wigner-Ville distribution (WVD) algorithms. Local binary pattern (LBP) texture features were extracted from the images, and Pearson correlation analysis identified sensitive features for distinguishing healthy and moldy-core pears. We built least squares support vector machine (LS-SVM) and extreme learning machine (ELM) classification models using these features, with CWT yielding the best results. The Sparrow Search Algorithm (SSA) optimized the models for both regions. Considering the spatial locality of moldy-core damage, we used D-S evidence theory to fuse classification results from the two sites and introduced a shrinkage–expansion function to improve accuracy. Results show that this approach increased the recognition capability of moldy-core by 2.56%. The classification accuracies were 94.87% for healthy pears and 97.92% for moldy pears, with an overall accuracy of 96.40%. This study offers a new method for high-precision non-destructive detection of internal defects in fruits and vegetables, which is valuable for improving quality control and sorting efficiency.</p> Graphical Abstract <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Novel Method of Detecting Pears with Moldy Core: Dual-Point Sensing of Acoustic Vibration and Decision-Level Fusion

  • 期刊论文

  • 发表日期：

    2026年4月1日

查看全文
  • Qinjun Zhao,
  • Jin Zhao,
  • Weiqi Yan,
  • Ye Song,
  • Tao Shen,
  • Kang Zhao

摘要

Moldy-core pear has no obvious symptoms in the early stages. The accuracy and reliability of single-point excitation and single-point sensing detection using acoustic vibration methods are low. To further enhance the detection accuracy of moldy-core disease in pears, we propose a new strategy using single-point excitation and dual-point sensing. This approach addresses challenges from internal tissue heterogeneity and anisotropic wave propagation in pears. We collected time-domain signals from the equatorial and calyx regions using self-built acoustic vibration devices. By integrating signals from different channels to enhance the accuracy and reliability of early moldy-core disease detection, the vibration response reflects the dynamic mechanical behavior of pear tissue, and internal moldy-core damage alters stiffness and damping characteristics, leading to changes in resonance behavior and time–frequency energy distribution. We transformed these signals into time–frequency images (TFIs) using short-time Fourier transform (STFT), continuous wavelet transform (CWT), and Wigner-Ville distribution (WVD) algorithms. Local binary pattern (LBP) texture features were extracted from the images, and Pearson correlation analysis identified sensitive features for distinguishing healthy and moldy-core pears. We built least squares support vector machine (LS-SVM) and extreme learning machine (ELM) classification models using these features, with CWT yielding the best results. The Sparrow Search Algorithm (SSA) optimized the models for both regions. Considering the spatial locality of moldy-core damage, we used D-S evidence theory to fuse classification results from the two sites and introduced a shrinkage–expansion function to improve accuracy. Results show that this approach increased the recognition capability of moldy-core by 2.56%. The classification accuracies were 94.87% for healthy pears and 97.92% for moldy pears, with an overall accuracy of 96.40%. This study offers a new method for high-precision non-destructive detection of internal defects in fruits and vegetables, which is valuable for improving quality control and sorting efficiency.

Graphical Abstract