Peningkatan Arsitektur Encoder-Decoder Modular Berbasis CNN pada Peningkatan Kualitas Citra Berkabut

Andika Muhammad; Arief Suryadi Satyawan; Mokh Mirza Etnisa Haqiqi; Fajar Rahmat Akbar; Iasya Faiqoh Nurrohmah; Aulia Adawiyah; Esti Fitria Wulandari; Rendi Tri Sugian; Nurul Fazri

doi:10.33005/santika.v6i1.914

Authors

Andika Muhammad Garut University
Arief Suryadi Satyawan Badan Riset dan Inovasi Nasional
Mokh Mirza Etnisa Haqiqi Teknik Elektro, Universitas Garut
Fajar Rahmat Akbar Teknik Elektro, Universitas Garut
Iasya Faiqoh Nurrohmah Teknik Elektro, Universitas Garut
Aulia Adawiyah Teknik Elektro, Universitas Garut
Esti Fitria Wulandari Teknik Elektro, Universitas Garut
Rendi Tri Sugian Teknik Elektro, Universitas Garut
Nurul Fazri Teknik Elektro, Universitas Garut

DOI:

https://doi.org/10.33005/santika.v6i1.914

Keywords:

Image Dehazing, CNN, Token-Mixing, Kendaraan Otonom

Abstract

Penelitian ini mengembangkan kerangka encoder–decoder berbasis CNN modular untuk tugas image dehazing, mengganti bottleneck konvensional dengan mekanisme token-mixing seperti FNet, Spatial-FNet, MLP-Mixer, dan gMLP. Pipeline mencakup pra-pemrosesan adaptif (CLAHE, histogram matching), augmentasi sintetik, serta pelatihan pada subset SOTS. Evaluasi numerik dan visual menunjukkan peningkatan signifikan dibanding baseline: rata-rata PSNR naik dari ≈18.4 dB menjadi ≈23.0–24.0 dB dan SSIM meningkat dari ≈0.75 menjadi ≈0.89–0.91. Temuan ini memberikan pedoman arsitektural dan strategi pra-pemrosesan bagi sistem visi dunia nyata seperti kendaraan otonom. Rekomendasi meliputi evaluasi pada dataset nyata lebih luas, tuning hiperparameter, analisis efisiensi komputasi, dan latensi sistem.

References

M. Shen, T. Lv, Y. Liu, J. Zhang, and M. Ju, “A Comprehensive Review of Traditional and Deep-Learning-Based Defogging Algorithms,” Electronics (Switzerland), vol. 13, no. 17, Sep. 2024, doi: 10.3390/electronics13173392.

P. V. Deshmukh, A. Kumar, and P. Chakrabarti, “Review of Deep Learning Based Image Dehazing for Autonomous Vehicle,” 2024.

C. Zheng, W. Ying, and Q. Hu, “Comparative analysis of dehazing algorithms on real-world hazy images,” Sci. Rep., vol. 15, no. 1, Dec. 2025, doi: 10.1038/s41598-025-95510-z.

J. Gui et al., “A Comprehensive Survey on Image Dehazing Based on Deep Learning,” 2021.

Kaiming He, Xiaoou Tang, and Jian Sun, Single image haze removal using dark channel prior, vol. 33. IEEE, 2011.

H. Ullah et al., “Light-DehazeNet: A Novel Lightweight CNN Architecture for Single Image Dehazing,” IEEE Transactions on Image Processing, vol. 30, pp. 8968–8982, 2021, doi: 10.1109/TIP.2021.3116790.

Z. Tu et al., “MAXIM: Multi-Axis MLP for Image Processing.” [Online]. Available: https://github.

A. Kholiq et al., “Enhancing Hazy Image Quality with a Modular CNN Encoder-Decoder,” ELKOMIKA, vol. 14, no. 1, pp. 69–83, Jan. 2026, doi: 10.26760/elkomika.

V. Stimper, S. Bauer, R. Ernstorfer, B. Schölkopf, and R. P. Xian, “Multidimensional Contrast Limited Adaptive Histogram Equalization,” IEEE Access, vol. 7, pp. 165437–165447, 2019, doi: 10.1109/ACCESS.2019.2952899.

Y. Liu, Q. Dong, C. Zhu, Y. Guo, and F. Wang, “Revitalizing image dehazing in the real world: A high-quality dataset and a customized method,” Comput. Vis. Media (Beijing)., vol. 11, no. 4, pp. 833–848, 2025, doi: 10.26599/CVM.2025.9450505.

H. Liu, Z. Dai, D. R. So, and Q. V. Le, “Pay Attention to MLPs,” Jun. 2021, [Online]. Available: http://arxiv.org/abs/2105.08050

Z. Chen, Z. He, and Z.-M. Lu, “DEA-Net: Single image dehazing based on detail-enhanced convolution and content-guided attention,” Jan. 2023, [Online]. Available: http://arxiv.org/abs/2301.04805

Tiande Mo, Wai-Yat Chan, Siqian Zheng, and Renhua Yang, “Review of AI Image Enhancement Techniques for In-Vehicle Vision Systems Under Adverse Weather Conditions,” World Electr. Veh. J, vol. 16, no. 2, Jan. 2025.

S. M. Sopian et al., “Development of a Modified CycleGAN Model with Residual Blocks and Perceptual Loss for Image Dehazing,” Teknika, vol. 14, no. 2, pp. 232–238, Jul. 2025, doi: 10.34148/teknika.v14i2.1235.

J. M. J. Valanarasu, R. Yasarla, and V. M. Patel, “TransWeather: Transformer-based Restoration of Images Degraded by Adverse Weather Conditions,” Jun. 2022, [Online]. Available: http://arxiv.org/abs/2111.14813

Y. Song, Z. He, H. Qian, and X. Du, “Vision Transformers for Single Image Dehazing,” Apr. 2022, doi: 10.1109/TIP.2023.3256763.

Y. Song, Z. He, H. Qian, and X. Du, “Vision Transformers for Single Image Dehazing,” IEEE Transactions on Image Processing, vol. PP, p. 1, Mar. 2023, doi: 10.1109/TIP.2023.3256763.

S. Satrasupalli, E. Daniel, and S. Guntur, “Single Image Haze Removal Based on Transmission Map Estimation Using Encoder-Decoder Based Deep Learning Architecture,” Optik (Stuttg)., vol. 248, p. 168197, Oct. 2021, doi: 10.1016/j.ijleo.2021.168197.

H. Zhou, Z. Chen, Y. Liu, Y. Sheng, W. Ren, and H. Xiong, “Physical-priors-guided DehazeFormer,” Knowl. Based. Syst., vol. 266, p. 110410, 2023, doi: https://doi.org/10.1016/j.knosys.2023.110410.

S. Okyere-Gyamfi, M. Asante, K. O. Peasah, Y. M. Missah, and V. Akoto-Adjepong, “Contrast limited adaptive histogram equalization (CLAHE) and colour difference histogram (CDH) feature merging capsule network (CCFMCapsNet) for complex image recognition,” PLoS One, vol. 20, no. 10 October, Oct. 2025, doi: 10.1371/journal.pone.0335393.