A Parallel Gaussian Filtering Algorithm Based on Color Difference

In order to remove noise effectively and reduce the loss of original information in image processing, a new parallel Gaussian filtering algorithm (PGF) based on Graphics Processing Units (GPU) is presented in this letter. The proposed method compares color difference between neighboring of the pixel and the center of the pixel before making Recursive Gaussian filter, moreover it can improve the speed of calculation using GPU. In our experiment, the image of LiNa (size of 512 pixel×512 pixel) is chosen to test our algorithm. The results show that our algorithm improves the PSNR of image, and the speed of GPU processing is more than eight times faster than CPU processing.