Performance Tuning for CUDA - Accelerated Neighborhood Denoising Filters
نویسنده
چکیده
Ziyi Zheng and Klaus Mueller are with the Center for Visual Computing, Computer Science, Stony Brook University, Stony Brook, NY 11790 USA (phone: 631-632-1524; e-mail: {zizhen, mueller}@cs.sunysb.edu). Funding was provided by NSF grant EAGER 1050477. Abstract—Neighborhood denoising filters are powerful techniques in image processing and can effectively enhance the image quality in CT reconstructions. In this study, by taking the bilateral filter and the non-local mean filter as two examples, we discuss their implementations and perform fine-tuning on the targeted GPU architecture. Experimental results show that the straightforward GPU-based neighborhood filters can be further accelerated by pre-fetching. The optimized GPU-accelerated denoising filters are ready for plug-in into reconstruction framework to enable fast denoising without compromising image quality.
منابع مشابه
A neighborhood regression approach for removing multiple types of noises
Image denoising is an important first step to provide cleaned images for follow-up tasks such as image segmentation and object recognition. Many image denoising filters have been proposed, with most of the filters focusing on one particular type of additive or multiplicative noise. In this article, we propose a novel neighborhood regression approach. Using the neighboring pixels as predictors, ...
متن کاملImage and movie denoising by nonlocal means
Neighborhood filters are image and movie filters which reduce the noise by averaging similar pixels. The object of the paper is to present a unified theory of these filters and reliable criteria to compare them to other classes of filters. First a CCD noise model will be presented justifying this class of algorithm. A classification of neighborhood filters is proposed, including classical image...
متن کاملComparing GPU Implementations of Bilateral and Anisotropic Diffusion Filters for 3D Biomedical Datasets
We compare the performance of hand-tuned CUDA implementations of bilateral and anisotropic diffusion filters for denoising 3D MRI datasets. Our tests sweep comparable parameters for the two filters and measure total runtime, memory bandwidth, computational throughput, and mean squared errors relative to a noiseless reference dataset.
متن کاملDiffusion Interpretation of Nonlocal Neighborhood Filters for Signal Denoising
Nonlocal neighborhood filters are modern and powerful techniques for image and signal denoising. In this paper, we give a probabilistic interpretation and analysis of the method viewed as a random walk on the patch space. We show that the method is intimately connected to the characteristics of diffusion processes, their escape times over potential barriers, and their spectral decomposition. In...
متن کاملOn a non-local spectrogram for denoising one-dimensional signals
In previous works, we investigated the use of local filters based on partial differential equations (PDE) to denoise one-dimensional signals through the image processing of timefrequency representations, such as the spectrogram. In this image denoising algorithms, the particularity of the image was hardly taken into account. We turn, in this paper, to study the performance of non-local filters,...
متن کامل