研究目的
Investigating the effectiveness of the non-local means method with Shapiro-Wilk test for sample selection in denoising images corrupted with additive Gaussian noise.
研究成果
The proposed non-local means filter using Shapiro-Wilk statistical test (SWNLM) demonstrates superior performance in denoising images corrupted with additive Gaussian noise, as validated by experiments on standard test images. The method achieves good noise reduction and detail preservation, outperforming other NLM-based methods in most cases.
研究不足
The proposed SWNLM filter has a high time complexity compared to conventional NLM due to the additional time taken for the Shapiro-Wilk test. A GPU implementation could reduce its execution time.
1：Experimental Design and Method Selection:
The study modifies the conventional non-local means filter (CNLM) by incorporating the Shapiro-Wilk test for sample selection to improve denoising performance.
2：Sample Selection and Data Sources:
Standard test images such as Cameraman, Lena, Boat, and Fingerprint images were used, artificially corrupted with additive white Gaussian noise (AWGN) of varying standard deviations.
3：List of Experimental Equipment and Materials:
The experiments were conducted using MATLAB 2018a on an Intel Core i7 CPU @
4：40GHz processor. Experimental Procedures and Operational Workflow:
The proposed method was compared with CNLM and other NLM-based methods in terms of PSNR, mean SSIM, and BC.
5：Data Analysis Methods:
Quantitative analysis was performed using peak signal-to-noise ratio (PSNR), mean structural similarity index matrix (SSIM), and Bhattacharya coefficient (BC).
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
