融合多层次特征增强与层级式分类的家电塑料识别

崇鹏豪; 郑云龙; 杨傲松; 郭梦慈; 李世峰

doi:10.11999/JEIT260084

融合多层次特征增强与层级式分类的家电塑料识别

doi: 10.11999/JEIT260084 cstr: 32379.14.JEIT260084

崇鹏豪¹,
郑云龙¹,
杨傲松¹,
郭梦慈¹,
李世峰^{1, 2, ,}

1.
河北经贸大学管理科学与信息工程学院石家庄 050061
2.
河北省跨境电商技术创新中心石家庄 050061

基金项目: 河北省教育厅科学研究项目(QN2024142)

详细信息

作者简介:
崇鹏豪：男，硕士生，研究方向为模式识别与人工智能

郑云龙：男，硕士生，研究方向为模式识别与人工智能、微波集成电路

杨傲松：男，硕士生，研究方向为计算机视觉与图像处理

郭梦慈：女，硕士生，研究方向为机器学习、自然语言处理

李世峰：男，讲师，研究方向为大数据与人工智能、微波集成电路

通讯作者:
李世峰　shifeng_li@hueb.edu.cn

中图分类号: TP391.41; TN911.73
计量
- 文章访问数: 41
- HTML全文浏览量: 8
- PDF下载量: 5
- 被引次数: 0
出版历程
- 收稿日期: 2026-01-22
- 修回日期: 2026-04-17
- 录用日期: 2026-04-17
- 网络出版日期: 2026-04-30

Household Appliance Plastics Identification by Fusing Multi-Level Feature Enhancement and Hierarchical Classification

1.
School of Management Science And Information Engineering, Hebei University of Economics and Business, Shijiazhuang 050061, China
2.
Hebei Provincial Cross-border E-commerce Technology Innovation Center, Hebei University of Economics and Business, Shijiazhuang 050061, China

Funds: Scientific Research Project of the Education Department of Hebei Province(QN2024142)

摘要

摘要: 针对废旧家电塑料回收中低光谱分辨率条件下识别精度不足的问题，尤其是黑色塑料在可见–近红外波段因高吸光性与光谱重叠导致的特征可分性下降，该文提出一种面向受限光谱特征空间的自动识别方法。该方法基于可见–近红外多光谱传感系统，通过多层次特征工程增强低维光谱信息，并结合分阶段层级分类策略，将多类别识别分解为粗分类与细分类的联合推理过程，以抑制复杂样本的特征混叠对分类性能的干扰。在五折交叉验证条件下，模型取得97.4%的分类准确率，在独立测试集上取得93.1%的分类准确率，相较于单阶段分类及无特征增强方法表现出更优的分类性能。结果表明，该方法在低分辨率多光谱条件下对黑色塑料及其他常见家电塑料具有稳定的识别能力，有效提升了复杂样本的分类精度，为废旧家电塑料自动化分选系统的开发提供了理论与应用支持。
- 塑料识别 /
- 可见–近红外光谱 /
- 层级聚类 /
- 特征增强
Abstract: Objective Accurate identification of plastics in waste household appliance recycling remains challenging under low-resolution spectral conditions. In practical recycling environments, plastics often exhibit complex compositions, surface contamination, and aging effects, which increase classification difficulty. In particular, black plastics show strong light absorption and spectral overlap in the visible and near infrared (Vis–NIR) range, leading to reduced feature separability and degraded classification performance. Under such conditions, conventional single-stage classification models are often unable to maintain stable accuracy. To address this problem, an automated identification method for low dimensional multispectral feature spaces is proposed, aiming to improve the discriminative capability of limited spectral information and enhance classification accuracy for complex plastic categories. Methods A compact Vis–NIR multispectral acquisition system based on the AS7265x sensor is used to collect 18 channel reflectance data within the 410–940 nm range. A handheld acquisition device with a controlled optical structure is designed to reduce environmental interference and ensure measurement consistency (Fig.3). A total of 576 samples from five typical appliance plastics, including ABS, high-impact polystyrene (HIPS), polypropylene (PP), acrylonitrile styrene copolymer (AS), and PC/ABS blends, are collected from waste household appliances and subjected to preliminary surface cleaning prior to spectral acquisition. To improve feature representation, a multi-level feature engineering strategy is adopted. It integrates original spectral intensity features, nonlinear polynomial expansion features, and adjacent channel ratio features to characterize both global and local spectral information. The nonlinear expansion enhances reflectance variation representation, while the ratio features capture local spectral shape changes and reduce external disturbances. These features are combined into a 53-dimensional feature vector. Linear Discriminant Analysis (LDA) is then applied to enhance inter class separability. To handle spectral overlap and class imbalance, a Hierarchical Joint Classifier (HJC) is constructed. The HJC adopts a two stage classification framework: an XGBoost based primary classifier performs coarse classification to separate easily distinguishable samples and group spectrally similar black plastics, while a TabTransformer based secondary classifier is used for fine grained classification of difficult samples (Fig. 6). This hierarchical design reduces classification complexity and improves discrimination for challenging categories. Model performance is evaluated using five fold cross validation and an independent test set. Evaluation metrics, including accuracy, precision, recall, and F1 score, are calculated based on confusion matrices (Fig. 7). Comparative experiments are conducted with traditional machine learning methods, ensemble learning models, and deep learning approaches under different feature processing strategies (Fig. 8, Fig. 9). Results and Discussions The proposed HJC achieves a classification accuracy of 97.4% under five fold cross validation and 93.1% on the independent test set (Table 4). Compared with single stage classifiers and methods without feature enhancement, the proposed method shows improved performance and stability under low resolution spectral conditions. Comparative results indicate that the proposed method outperforms baseline approaches such as PCA combined with CNN, which achieves approximately 71.3% accuracy under the same dataset (Fig. 8). This improvement indicates that the proposed feature engineering strategy effectively enhances the discriminative capability of low dimensional spectral data. In addition, combining LDA with feature engineering further improves class separability compared with conventional PCA based methods. Confusion matrix analysis shows that misclassifications are mainly concentrated between spectrally similar black ABS and black HIPS samples, while other categories achieve high classification accuracy (Fig. 9). This result suggests that spectral overlap remains the main challenge under low resolution conditions. The hierarchical classification strategy mitigates this issue by focusing classification on difficult samples, thereby improving the overall generalization capability of the model. Overall, the proposed method demonstrates robustness under practical conditions, including spectral noise, limited channel resolution, and material heterogeneity, indicating its suitability for real world recycling applications. Conclusions A hierarchical classification method with multi-level spectral feature engineering is developed for plastic identification under low-resolution Vis–NIR conditions. Nonlinear and morphological features are incorporated into a two-stage framework to improve performance on spectrally similar materials. The results show consistent accuracy across different plastic types. The method is suitable for automated sorting in waste appliance recycling and can be applied to other material identification tasks with limited spectral information.
- Plastic classification /
- Visible–near infrared spectroscopy /
- Hierarchical classification /
- Feature enhancement

HTML全文

图 1 可见–近红外光谱技术原理示意图

下载: 全尺寸图片幻灯片

图 2 低成本光谱传感器实物图

下载: 全尺寸图片幻灯片

图 3 数据采集装置结构及工作示意图

下载: 全尺寸图片幻灯片

图 4 不同塑料样本在可见–近红外波段的多光谱反射响应对比

下载: 全尺寸图片幻灯片

图 5 模型训练、部署、推理流程图

下载: 全尺寸图片幻灯片

图 6 分层级联合分类器模型(HJC)结构图

下载: 全尺寸图片幻灯片

图 7 二元分类混淆矩阵示意图

下载: 全尺寸图片幻灯片

图 8 不同模型在多种特征表示条件下的分类精度对比

下载: 全尺寸图片幻灯片

图 9 LDA结合特征工程各模型混淆矩阵

下载: 全尺寸图片幻灯片

表 1 三个子芯片对应的6通道波长说明

子芯片	波段类型	6 个通道中心波长(nm)
AS72651	近红外(VIS–NIR)	610、680、730、760、810、860
AS72652	可见–近红外(VIS–NIR)	560、585、645、705、900、940
AS72653	可见光(VIS)	410、435、460、485、510、535

下载: 导出CSV

表 2 不同参数探测头参数以及效果说明

探测头编码	类型	内径(mm)	高度(mm)	$ {\text{SD}}_{\text{mean}} $	$ {\text{Delta}}_{\text{mean}} $
1	圆柱形	25	45	1.18949	2.979367
2	圆柱形	25	50	1.33102	3.267261
3	圆柱形	18	40	1.612278	3.736211
4	圆柱形	20	35	1.977449	5.234972
5	圆柱形	20	22	0.967459	2.218467
6	圆锥形	10/20	35	1.197082	2.911761
7	圆锥形	10/20	30	1.343622	3.401544

下载: 导出CSV

表 3 LDA、特征工程条件HJC模型评估指标

塑料类别	精确率	召回率	F1-Score
ABS-BLACK	0.90	0.94	0.92
ABS-WHITE	1.00	1.00	1.00
AS-BLACK	0.99	1.00	0.99
AS-WHITE	1.00	1.00	1.00
HIPS-BLACK	0.92	0.87	0.89
HIPS-WHITE	1.00	1.00	1.00
PC+ABS	1.00	1.00	0.99
PP	1.00	1.00	1.00

下载: 导出CSV

表 4 独立测试集上不同模型分类性能及排序结果

排名	分类器	测试准确率	宏平均F1-Score	加权平均F1-Score	测试样本数
1	HJC(This Work)	0.9310	0.9202	0.9307	145
2	XGBoost	0.9103	0.8915	0.9133	145
3	ResNet	0.8966	0.8805	0.8998	145
4	TabTransformer	0.8759	0.8590	0.8752	145
5	Logistic Regression	0.8690	0.8679	0.8825	145
6	Stacking	0.8552	0.8471	0.8615	145
7	SVM	0.8552	0.8511	0.8617	145
8	Random Forest	0.8552	0.8591	0.8600	145
9	CNN	0.8207	0.8056	0.8224	145

下载: 导出CSV

表 5 HJC模型在独立测试集上的分类别性能指标

塑料类别	精确率	召回率	F1-Score	样本数
ABS−BLACK	0.9688	0.9143	0.9412	35
ABS−WHITE	1.0000	0.9600	0.9796	25
AS−BLACK	1.0000	0.7273	0.8421	11
AS−WHITE	1.0000	1.0000	1.0000	12
HIPS−BLACK	0.6471	0.8571	0.7273	14
HIPS−WHITE	0.8667	1.0000	0.9286	13
PC+ABS	0.9444	1.0000	0.9714	17
PP	1.0000	0.9444	0.9714	18

下载: 导出CSV

表 6 不同传感器条件下塑料识别方法的性能与成本对比

参考文献	传感器	可识别塑料类型	准确率	实验建模方法	系统成本
[8]	S6000-FTIR	ABS, PS, PP, PE, PET, PVC	100%	PCA+FDA	高
[12]	彩谱FS-15	PET,PE,PVC,PP,PS,PC,POM,ABS	98.67%	GA+SVM	中
[14]	SpectraPod	PET, PVC,PP,PS,PE	93%	SVM	中
[15]	AS7265x	PET,PVC,PP,PS,PE	72.5%	PCA+CNN	低
本工作	AS7265x	ABS,HIPS,PP,AS,PC+ABS混合材料	93.1%	LDA+特征工程+HJC	低

下载: 导出CSV

参考文献(20)

[1]	Organisation for Economic Co-operation and Development (OECD). Global plastics outlook: Policy scenarios to 2060[R]. 2022. (查阅网上资料, 未找到本条文献的报告编号, 请核对).
[2]	United Nations Environment Programme (UNEP). From pollution to solution: A global assessment of marine litter and plastic pollution[R]. 2021. (查阅网上资料, 未找到本条文献的报告编号, 请核对).
[3]	蔡毅, 田晖, 谢淼雪. 废旧家用电器塑料资源化利用及发展趋势[J]. 中国塑料, 2021, 35(8): 77–83. doi: 10.19491/j.issn.1001-9278.2021.08.010. CAI Yi, TIAN Hui, and XIE Miaoxue. Utilization and development trend of waste plastics from household electrical appliances as resources[J]. China Plastics, 2021, 35(8): 77–83. doi: 10.19491/j.issn.1001-9278.2021.08.010.
[4]	ZHENG Xuqing, DENG Mike, JIA Hao, et al. Surface modification of intumescent flame retardant and its application in polypropylene with excellent fire performance and water resistance[J]. Polymers, 2025, 17(3): 399. doi: 10.3390/polym17030399.
[5]	DENG Mike, ZHANG Zhihao, SUN Jun, et al. Improving the flame retardancy and water resistance of polypropylene by introducing microcapsule flame retardant system and modified zinc oxide[J]. Polymer Degradation and Stability, 2024, 221: 110668. doi: 10.1016/j.polymdegradstab.2024.110668.
[6]	DE OLIVEIRA SILVA R J, GRAF K, and LEITE RIBEIRO OKIMOTO M L. Plastic waste recycling: An overview of the mechanical, chemical, and thermal technologies[J]. Journal of Engineering and Applied Science, 2025, 72(1): 251. doi: 10.1186/s44147-025-00799-2.
[7]	LUBONGO C, BIN DAEJ M A A, and ALEXANDRIDIS P. Recent developments in technology for sorting plastic for recycling: The emergence of artificial intelligence and the rise of the robots[J]. Recycling, 2024, 9(4): 59. doi: 10.3390/recycling9040059.
[8]	ZHENG Yan, BAI Jiarui, XU Jingna, et al. A discrimination model in waste plastics sorting using NIR hyperspectral imaging system[J]. Waste Management, 2018, 72: 87–98. doi: 10.1016/j.wasman.2017.10.015.
[9]	KISSEL A, NOGOWSKI A, KIENLE A, et al. Flow Raman spectroscopy for the detection and identification of small microplastics[J]. Sensors, 2025, 25(5): 1390. doi: 10.3390/s25051390.
[10]	VÁZQUEZ-GUARDADO A, MONEY M, MCKINNEY N, et al. Multi-spectral infrared spectroscopy for robust plastic identification[J]. Applied Optics, 2015, 54(24): 7396–7405. doi: 10.1364/AO.54.007396.
[11]	胡锡敦, 尹禄, 杨钦晨, 等. 基于近红外高光谱成像技术的塑料分类(特邀)[J]. 激光与光电子学进展, 2024, 61(2): 0211031. doi: 10.3788/LOP232402. HU Xidun, YIN Lu, YANG Qinchen, et al. Classification of plastics based on near-infrared hyperspectral imaging technology (Invited)[J]. Laser & Optoelectronics Progress, 2024, 61(2): 0211031. doi: 10.3788/LOP232402.
[12]	SINGH A R, NEO E R K, LAI C M, et al. Deep learning-based plastic classification using spectroscopic data[J]. Journal of Cleaner Production, 2025, 530: 146793. doi: 10.1016/j.jclepro.2025.146793.
[13]	赵凤, 耿苗苗, 刘汉强, 等. 卷积神经网络与视觉Transformer联合驱动的跨层多尺度融合网络高光谱图像分类方法[J]. 电子与信息学报, 2024, 46(5): 2237–2248. doi: 10.11999/JEIT231209. ZHAO Feng, GENG Miaomiao, LIU Hanqiang, et al. Convolutional neural network and vision transformer-driven cross-layer multi-scale fusion network for hyperspectral image classification[J]. Journal of Electronics & Information Technology, 2024, 46(5): 2237–2248. doi: 10.11999/JEIT231209.
[14]	VAN HOORN H, POURMOHAMMADI F, DE LEEUW A W, et al. Machine learning-based identification of plastic types using handheld spectrometers[J]. Sensors, 2025, 25(12): 3777. doi: 10.3390/s25123777.
[15]	MARTINEZ-HERNANDEZ U, WEST G, and ASSAF T. Low-cost recognition of plastic waste using deep learning and a multi-spectral near-infrared sensor[J]. Sensors, 2024, 24(9): 2821. doi: 10.3390/s24092821.
[16]	RANI M, MARCHESI C, FEDERICI S, et al. Miniaturized near-infrared (MicroNIR) spectrometer in plastic waste sorting[J]. Materials, 2019, 12(17): 2740. doi: 10.3390/ma12172740.
[17]	蔡轶珩, 谭美伶, 潘建军, 等. 基于多尺度非对称密集网络的高光谱图像分类[J]. 电子与信息学报, 2024, 46(4): 1448–1457. doi: 10.11999/JEIT230651. CAI Yiheng, TAN Meiling, PAN Jianjun, et al. Hyperspectral image classification based on multi-scale asymmetric dense network[J]. Journal of Electronics & Information Technology, 2024, 46(4): 1448–1457. doi: 10.11999/JEIT230651.
[18]	刘晓敏, 余梦君, 乔振壮, 等. 面向多源遥感数据分类的尺度自适应融合网络[J]. 电子与信息学报, 2024, 46(9): 3693–3702. doi: 10.11999/JEIT240178. LIU Xiaomin, YU Mengjun, QIAO Zhenzhuang, et al. Scale adaptive fusion network for multimodal remote sensing data classification[J]. Journal of Electronics & Information Technology, 2024, 46(9): 3693–3702. doi: 10.11999/JEIT240178.
[19]	肖振久, 田昊, 张杰浩, 等. 融合动态特征增强的遥感建筑物分割[J]. 光电工程, 2025, 52(3): 240231. doi: 10.12086/oee.2025.240231. XIAO Zhenjiu, TIAN Hao, ZHANG Jiehao, et al. Fusion of dynamic features enhances remote sensing building segmentation[J]. Opto-Electronic Engineering, 2025, 52(3): 240231. doi: 10.12086/oee.2025.240231.
[20]	吕鹏远, 兰金江, 曾学仁, 等. 基于特征增强与融合的红外目标检测算法[J]. 红外技术, 2024, 46(7): 782–790. LÜ Pengyuan, LAN Jinjiang, ZENG Xueren, et al. Infrared object detection algorithm based on feature enhancement and fusion[J]. Infrared Technology, 2024, 46(7): 782–790.