A single-wavelength laser relaxation spectroscopy-based machine learning solution for apple mechanical damage detection

Detecting and mitigating mechanical damage in apples during picking and transportation is a critical concern in the agricultural industry. This paper investigates the optimization of a pattern recognition model using single-wavelength laser relaxation spectroscopy for the purpose of apple mechanical...

Full description

Saved in:
Bibliographic Details
Published in:Multimedia tools and applications 2024-01, Vol.83 (24), p.64617-64635
Main Authors: Lian, Junbo, Zhang, Jingyu, Liu, Quan, Zhu, Runhao, Ning, Jingyuan, Xiong, Siyi, Hui, Guohua, Gao, Yuanyuan, Lou, Xiongwei
Format: Article
Language:English
Subjects:
Accuracy
Apples
Classification
Computer Communication Networks
Computer Science
Cost analysis
Cost effectiveness
Damage detection
Data Structures and Information Theory
Lasers
Machine learning
Multimedia Information Systems
Pattern recognition
Search algorithms
Special Purpose and Application-Based Systems
Spectroscopy
Spectrum analysis
Support vector machines
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Detecting and mitigating mechanical damage in apples during picking and transportation is a critical concern in the agricultural industry. This paper investigates the optimization of a pattern recognition model using single-wavelength laser relaxation spectroscopy for the purpose of apple mechanical damage detection. We conducted experiments using Red Fuji apples as our sample dataset and designed a single-wavelength laser relaxation spectroscopy system to collect spectral data. The 1823 sets of data from 45 apples were collected. To enhance the quality of the data, we applied the Min–Max standardization algorithm for preprocessing. Subsequently, we employed multiple pattern recognition models, including Support Vector Machine (SVM), Cross-Validation Optimized Support Vector Machine (CV-SVM), Relevance Vector Machine (RVM), and Sparrow Search Algorithm Optimized Relevance Vector Machine (SSA-RVM), to establish models for apple damage detection. Our study involved a comparison of the efficiency and accuracy of these models. Our findings indicate that CV-SVM emerged as the most stable model for apple damage detection, achieving an impressive accuracy rate of 93.19%. Furthermore, to enhance the classification performance, we applied Multiple Measurement Classification Recognition (MMCR). The CV-SVM-MMCR model demonstrated superior classification abilities, with a detection accuracy rate of 97.5%. Notably, our proposed method offers several advantages, including ease of operation, rapid analysis, and cost-effectiveness.
ISSN:1573-7721
1380-7501
1573-7721
DOI:10.1007/s11042-023-18038-2