CUDA-based quick bilateral filtering method

Abstract

The invention discloses a CUDA-based quick bilateral filter method. The method mainly overcomes the defect of a standard bilateral filtering algorithm that computation speed is low. According to the method, parallelization is conducted on the original bilateral filtering algorithm, then image data are copied from a host memory into a GPU memory, a GPU (Graphic Processing Unit) is called through a CUDA (Computer Unified Device Architecture) for filtering computation, and parallel execution is adopted for computation of all pixel points. Compared with the existing bilateral filtering algorithm, the CUDA-based quick bilateral filter method has the advantage that on the premise that noise reduction performance is maintained, computing speed is increased greatly, so that the requirement for real-time processing is met.

Description

technical field [0001] The invention relates to the field of graphic image processing and parallel computing, in particular to a fast bilateral filtering method based on CUDA. Background technique [0002] The image sensor will be affected by various factors during the working process, such as the environment in which the image is captured, the quality of the sensor components themselves, etc. Therefore, the digital image will be polluted by various noises during the formation process. These noises will have a great impact on image post-processing. How to filter out these noises while maintaining the original information of the image is very important. Traditional filters also blur the edges of the image while removing noise. To solve this problem, Tomasi and Manduchi proposed bilateral filtering algorithm in 1998. Similar to Gaussian filtering, bilateral filtering also adopts the idea of ​​local weighted average, the difference is that bilateral filtering adds pixel valu...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a fast bilateral filtering method based on CUDA, aiming to solve the problem that the complexity of the bilateral filtering algorithm is high and cannot meet the requirements of real-time image processing

Images