Image RestorationComp344TutorialKai Zhang
Image Restoration Comp344 Tutorial Kai Zhang
OutlineDIPUM Tool boxNoisemodelsPeriodical noise and removalNoiseparameter estimationSpatiral noise removal
Outline ◼ DIPUM Tool box ◼ Noise models ◼ Periodical noise and removal ◼ Noise parameter estimation ◼ Spatiral noise removal
The DIPUM Tool boxM-functions from the book Digital ImageProcessingUsingMATLABFreedownloadNo original m-functionsButcan still usedfordemo
The DIPUM Tool box ◼ M-functions from the book Digital Image Processing Using MATLAB ◼ Freedownload ◼ No original m-functions ◼ But can still used for demo
Noise modelsGiven a random number generator,how togeneraterandom numbers with a pre-specified CDF?Suppose the random number w is in [0,1]We want to generate a Rayleigh distributed sample set(1-e-(=-a) b,z>=aF(2) = (0,<aTofindz solvethefollowing equation-(z-a)2 / b1-e=Wz = a + /-bln(l-w)
Noise models ◼ Given a random number generator, how to generate random numbers with a pre-specified CDF? ◼ Suppose the random number w is in [0,1] ◼ We want to generate a Rayleigh distributed sample set ◼ To find z solve the following equation e z a z a z a b F z − = − − = 1 , 0, / 2 ( ) ( ) ln(1 ) 1 / 2 ( ) z a b w e w z a b = + − − − = − −
FunctionsFunction r = imnoise(f, type, parameters)Corrupt image f with noise specified in typeand parametersResults returned inrType include: uniform, gaussian, salt &pepper, lognormal, rayleigh, exponential Function r = imnoise2(type, M,N,a,b);Generates arrar r of size M-by-N,Entries are of the specified distribution typeA and b are parameters
Functions ◼ Function r = imnoise(f, type, parameters) ◼ Corrupt image f with noise specified in type and parameters ◼ Results returned in r ◼ Type include: uniform, gaussian, salt & pepper, lognormal, rayleigh, exponential ◼ Function r = imnoise2(type, M,N,a,b); ◼ Generates arrar r of size M-by-N, ◼ Entries are of the specified distribution type ◼ A and b are parameters