Abstract:

The MRI (magnetic resonance imaging) image denoising is an important step in image preprocessing.An improved neighborhood average method is proposed to compensate for the inability of average filtering algorithms to retain image edges and details in the filtering.Firstly,the relationship between the pixel and each template is determined according to the compatibility and similitude principle to judge whether the pixel is an isolated noise point.Then the isolated noise points are processed by the similar gray T neighborhood average method,and the non isolated ones using weighted smoothing template.The experimental result shows that the improved algorithm retains the image edges and details while removing the noise.

Key words: image denoising;neighborhood average method;improved algorithm