Date of Publication :7th March 2015

Abstract: The future of computation is the GPU, i.e. Graphical Processing Unit. They are developing into great parallel computing units with the promise that the graphics cards have shown in the field of image processing and the computational capability that these GPUs possess. Compute Unified Device Architecture i.e. CUDA is NVIDIA’s parallel computing architecture. It enables dramatic increase in computing performance, by completely harnessing the power of the GPU. Initially we generate a MATLAB code for image processing using bilateral filter, and then we interface GPU with CPU using CUDA and create the related code of image processing for GPU for this filter. By changing the values of distinct parameters of the images, we have obtained variations in the output images and their corresponding computational time on CPU as well as on GPU. Further these computational times are compared and it is observed that the time taken by the GPU is 70 to 80% less than that of CPU.

Reference :