Enhancing Low-Light Images with Kolmogorov–Arnold Networks in Transformer Attention
Low-light image enhancement (LLIE) techniques improve the performance of image sensors by enhancing visibility and details in poorly lit environments and have significantly benefited from recent research into Transformer models. This work presents a novel Transformer attention mechanism inspired by...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-01-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/25/2/327 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832587583800999936 |
|---|---|
| author | Alexandru Brateanu Raul Balmez Ciprian Orhei Cosmin Ancuti Codruta Ancuti |
| author_facet | Alexandru Brateanu Raul Balmez Ciprian Orhei Cosmin Ancuti Codruta Ancuti |
| author_sort | Alexandru Brateanu |
| collection | DOAJ |
| description | Low-light image enhancement (LLIE) techniques improve the performance of image sensors by enhancing visibility and details in poorly lit environments and have significantly benefited from recent research into Transformer models. This work presents a novel Transformer attention mechanism inspired by the Kolmogorov–Arnold representation theorem, incorporating learnable non-linearity and multivariate function decomposition. This innovative mechanism is the foundation of KAN-<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi mathvariant="script">T</mi></semantics></math></inline-formula>, our proposed Transformer network. By enhancing feature flexibility and enabling the model to capture broader contextual information, KAN-<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi mathvariant="script">T</mi></semantics></math></inline-formula> achieves superior performance. Our comprehensive experiments, both quantitative and qualitative, demonstrate that the proposed method achieves state-of-the-art performance in low-light image enhancement, highlighting its effectiveness and wide-ranging applicability. The code will be released upon publication. |
| format | Article |
| id | doaj-art-93fbeda44bbe4e9c8dffd25c018dbd30 |
| institution | Kabale University |
| issn | 1424-8220 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj-art-93fbeda44bbe4e9c8dffd25c018dbd302025-01-24T13:48:31ZengMDPI AGSensors1424-82202025-01-0125232710.3390/s25020327Enhancing Low-Light Images with Kolmogorov–Arnold Networks in Transformer AttentionAlexandru Brateanu0Raul Balmez1Ciprian Orhei2Cosmin Ancuti3Codruta Ancuti4Department of Computer Science, University of Machester, Manchester M13 9PL, UKDepartment of Computer Science, University of Machester, Manchester M13 9PL, UKFaculty of Electronics, Telecommunications and Information Technologies, Polytechnic University Timisoara, 300223 Timisoara, RomaniaFaculty of Electronics, Telecommunications and Information Technologies, Polytechnic University Timisoara, 300223 Timisoara, RomaniaFaculty of Electronics, Telecommunications and Information Technologies, Polytechnic University Timisoara, 300223 Timisoara, RomaniaLow-light image enhancement (LLIE) techniques improve the performance of image sensors by enhancing visibility and details in poorly lit environments and have significantly benefited from recent research into Transformer models. This work presents a novel Transformer attention mechanism inspired by the Kolmogorov–Arnold representation theorem, incorporating learnable non-linearity and multivariate function decomposition. This innovative mechanism is the foundation of KAN-<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi mathvariant="script">T</mi></semantics></math></inline-formula>, our proposed Transformer network. By enhancing feature flexibility and enabling the model to capture broader contextual information, KAN-<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi mathvariant="script">T</mi></semantics></math></inline-formula> achieves superior performance. Our comprehensive experiments, both quantitative and qualitative, demonstrate that the proposed method achieves state-of-the-art performance in low-light image enhancement, highlighting its effectiveness and wide-ranging applicability. The code will be released upon publication.https://www.mdpi.com/1424-8220/25/2/327image sensor restorationlow-light enhancementVision Transformer |
| spellingShingle | Alexandru Brateanu Raul Balmez Ciprian Orhei Cosmin Ancuti Codruta Ancuti Enhancing Low-Light Images with Kolmogorov–Arnold Networks in Transformer Attention Sensors image sensor restoration low-light enhancement Vision Transformer |
| title | Enhancing Low-Light Images with Kolmogorov–Arnold Networks in Transformer Attention |
| title_full | Enhancing Low-Light Images with Kolmogorov–Arnold Networks in Transformer Attention |
| title_fullStr | Enhancing Low-Light Images with Kolmogorov–Arnold Networks in Transformer Attention |
| title_full_unstemmed | Enhancing Low-Light Images with Kolmogorov–Arnold Networks in Transformer Attention |
| title_short | Enhancing Low-Light Images with Kolmogorov–Arnold Networks in Transformer Attention |
| title_sort | enhancing low light images with kolmogorov arnold networks in transformer attention |
| topic | image sensor restoration low-light enhancement Vision Transformer |
| url | https://www.mdpi.com/1424-8220/25/2/327 |
| work_keys_str_mv | AT alexandrubrateanu enhancinglowlightimageswithkolmogorovarnoldnetworksintransformerattention AT raulbalmez enhancinglowlightimageswithkolmogorovarnoldnetworksintransformerattention AT ciprianorhei enhancinglowlightimageswithkolmogorovarnoldnetworksintransformerattention AT cosminancuti enhancinglowlightimageswithkolmogorovarnoldnetworksintransformerattention AT codrutaancuti enhancinglowlightimageswithkolmogorovarnoldnetworksintransformerattention |