Biorthogonal Lattice Tunable Wavelet Units and Their Implementation in Convolutional Neural Networks for Computer Vision Problems

This work introduces a universal wavelet unit constructed with a biorthogonal lattice structure which is a novel tunable wavelet unit to enhance image classification and anomaly detection in convolutional neural networks by reducing information loss during pooling. The unit employs a biorthogonal la...

Full description

Saved in:
Bibliographic Details
Main Authors: An D. Le, Shiwei Jin, Sungbal Seo, You-Suk Bae, Truong Q. Nguyen
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Open Journal of Signal Processing
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11039659/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849318677016477696
author An D. Le
Shiwei Jin
Sungbal Seo
You-Suk Bae
Truong Q. Nguyen
author_facet An D. Le
Shiwei Jin
Sungbal Seo
You-Suk Bae
Truong Q. Nguyen
author_sort An D. Le
collection DOAJ
description This work introduces a universal wavelet unit constructed with a biorthogonal lattice structure which is a novel tunable wavelet unit to enhance image classification and anomaly detection in convolutional neural networks by reducing information loss during pooling. The unit employs a biorthogonal lattice structure to modify convolution, pooling, and down-sampling operations. Implemented in residual neural networks with 18 layers, it improved detection accuracy on CIFAR10 (by 2.67% ), ImageNet1K (by 1.85% ), and the Describable Textures dataset (by 11.81% ), showcasing its advantages in detecting detailed features. Similar gains are achieved in the implementations for residual neural networks with 34 layers and 50 layers. For anomaly detection on the MVTec Anomaly Detection and TUKPCB datasets, the proposed method achieved a competitive performance and better anomaly localization.
format Article
id doaj-art-08be3afbe7b848a3824aed6ffc28b1af
institution Kabale University
issn 2644-1322
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Open Journal of Signal Processing
spelling doaj-art-08be3afbe7b848a3824aed6ffc28b1af2025-08-20T03:50:45ZengIEEEIEEE Open Journal of Signal Processing2644-13222025-01-01676878310.1109/OJSP.2025.358096711039659Biorthogonal Lattice Tunable Wavelet Units and Their Implementation in Convolutional Neural Networks for Computer Vision ProblemsAn D. Le0https://orcid.org/0009-0000-4684-715XShiwei Jin1Sungbal Seo2https://orcid.org/0009-0000-4226-0327You-Suk Bae3Truong Q. Nguyen4Electrical and Computer Engineering Department, University of California San Diego, La Jolla, CA, USAElectrical and Computer Engineering Department, University of California San Diego, La Jolla, CA, USADepartment of Computer Engineering, Tech University of Korea, Siheung, South KoreaDepartment of Computer Engineering, Tech University of Korea, Siheung, South KoreaElectrical and Computer Engineering Department, University of California San Diego, La Jolla, CA, USAThis work introduces a universal wavelet unit constructed with a biorthogonal lattice structure which is a novel tunable wavelet unit to enhance image classification and anomaly detection in convolutional neural networks by reducing information loss during pooling. The unit employs a biorthogonal lattice structure to modify convolution, pooling, and down-sampling operations. Implemented in residual neural networks with 18 layers, it improved detection accuracy on CIFAR10 (by 2.67% ), ImageNet1K (by 1.85% ), and the Describable Textures dataset (by 11.81% ), showcasing its advantages in detecting detailed features. Similar gains are achieved in the implementations for residual neural networks with 34 layers and 50 layers. For anomaly detection on the MVTec Anomaly Detection and TUKPCB datasets, the proposed method achieved a competitive performance and better anomaly localization.https://ieeexplore.ieee.org/document/11039659/Anomaly detectioncomputer visiondiscrete wavelet transformsfeature extractionimage processingimage recognition
spellingShingle An D. Le
Shiwei Jin
Sungbal Seo
You-Suk Bae
Truong Q. Nguyen
Biorthogonal Lattice Tunable Wavelet Units and Their Implementation in Convolutional Neural Networks for Computer Vision Problems
IEEE Open Journal of Signal Processing
Anomaly detection
computer vision
discrete wavelet transforms
feature extraction
image processing
image recognition
title Biorthogonal Lattice Tunable Wavelet Units and Their Implementation in Convolutional Neural Networks for Computer Vision Problems
title_full Biorthogonal Lattice Tunable Wavelet Units and Their Implementation in Convolutional Neural Networks for Computer Vision Problems
title_fullStr Biorthogonal Lattice Tunable Wavelet Units and Their Implementation in Convolutional Neural Networks for Computer Vision Problems
title_full_unstemmed Biorthogonal Lattice Tunable Wavelet Units and Their Implementation in Convolutional Neural Networks for Computer Vision Problems
title_short Biorthogonal Lattice Tunable Wavelet Units and Their Implementation in Convolutional Neural Networks for Computer Vision Problems
title_sort biorthogonal lattice tunable wavelet units and their implementation in convolutional neural networks for computer vision problems
topic Anomaly detection
computer vision
discrete wavelet transforms
feature extraction
image processing
image recognition
url https://ieeexplore.ieee.org/document/11039659/
work_keys_str_mv AT andle biorthogonallatticetunablewaveletunitsandtheirimplementationinconvolutionalneuralnetworksforcomputervisionproblems
AT shiweijin biorthogonallatticetunablewaveletunitsandtheirimplementationinconvolutionalneuralnetworksforcomputervisionproblems
AT sungbalseo biorthogonallatticetunablewaveletunitsandtheirimplementationinconvolutionalneuralnetworksforcomputervisionproblems
AT yousukbae biorthogonallatticetunablewaveletunitsandtheirimplementationinconvolutionalneuralnetworksforcomputervisionproblems
AT truongqnguyen biorthogonallatticetunablewaveletunitsandtheirimplementationinconvolutionalneuralnetworksforcomputervisionproblems