Image Denoising Method Based on 3D Block Matching with Harmonic Filtering in Transform Domain

Mizanur Rashid (Nanjing University of Information Science and Technology, China)
Abdullah Ibne Sayed (Huazhong University of Science and Technology, Wuhan, China)
Md Masud Rana (Huazhong University of Science and Technology, Wuhan, China)

International Journal of Informatics, Information System and Computer Engineering (INJIISCOM)
Vol. 7 No. 1 (2026): INJIISCOM: VOLUME 7, ISSUE 1, JUNE 2026 (ONLINE FIRST)

Digital images are highly susceptible to noise during acquisition and transmission, requiring efficient preprocessing methods. Traditional denoising techniques often suffer from high computational complexity. To address this challenge, this paper presents a novel and efficient denoising algorithm that integrates wave-domain harmonic filtering with 3D block matching (BM3D). Similar 2D image blocks are grouped into a 3D array using the Euclidean distance approach for joint filtering. Following inverse transformation, wavelet decomposition isolates high-frequency noise. To prevent edge blurring and distortion, a Laplacian-Gaussian algorithm is incorporated to refine the diffusion model. Experimental results demonstrate that the proposed model significantly improves information protection, edge preservation, and computational speed