|
计算机应用研究 2009
Image median filtering algorithm based on impulse noise detection
|
Abstract:
The paper proposed an algorithm that removes the impulse noise in an image by median filtering on the basis of the impulse noise detection result.It first mapped the sub image sample space mixing with impulse noise into a supersphere of multidimensions by kernel function,then computed the radius R of the sphere and the vector of sphere center a.Compared the radius R with the distance that the test sample and sphere center a.If the absolute value of them subtracting was less than a threshold,then inexist noise, contrarily exist noise.The median filtering was used to remove the noise. Compareing with the others algorithms,it could verdict accurately the impulse noise much more,the mode filtering noise was more reasonable,the images range applicability was morebroad,so,it holded better filtering performance.