Method for noise reduction in images in an image sequence

A method for noise suppression in images of an image sequence in accordance with the following algorithm:

• Pi,jn+1=αi,jn+1*Pi,jn+(1−αi,jn+1)*Qi,jn+1
• where:
• Pi,jn+1=grayscale value of a pixel of the ith row and the jth column after the (n+1)th iteration,
• Pi,jn=grayscale value of a pixel of the ith row and the jth column after the (n)th iteration,
• Qi,jn+1=grayscale value of a pixel of the ith row and the jth column after the (n+1)th recording (unfiltered), and
• di,jn+1=attenuation function for a pixel in the ith row and the jth column,
• the algorithm being a recursive law-pass with an attenuation function αi,jn+1 adapted for the pixel, and the action of the attenuation being dependent on a grayscale value difference dQi,jn+1=|Pi,jn−Qi,jn+1| between. filtered and unfiltered pixels, on an amplitude of the grayscale value Qi,jn+1 of the pixel, it further being the case that:
• γ
• i
• ,
• j
• n
• +
• 1
• 
• {
• =
• Q
• 
• 
• 0
• Q
• i
• ,
• j
• n
• +
• 1
• [
• for
• 
• 
• Q
• i
• ,
• j
• n
• +
• 1
• >
• Q
• 
• 
• 0
• ]
• =
• 1
• [
• for
• 
• 
• otherwise
• ]
• where γi,jn+1=weighting factor, and
• Q0—reference grayscale value,
• characterised in that
• an adaptation of the strength of the attenuation factor α0i,jn+1 is performed taking account of the sequence length in which a pixel of the ith row and the jth column has an unchanged grayscale value,
• the attenuation factors α0i,jn+1 with a limiting value α0min and a limiting value α0max are defined,
• a movement measure
• Γ
• i
• ,
• j
• n
• +
• 1
• 
• {
• =
• 
• Q
• i
• ,
• j
• n
• +
• 1
• dif_swl
• *
• γ
• i
• ,
• j
• n
• +
• 1
• [
• for
• 
• 
• 
• Q
• i
• ,
• j
• n
• +
• 1
• *
• γ
• i
• ,
• j
• n
• +
• 1
• <
• dif_swl
• ]
• =
• 1
• [
• otherwise
• ]
• is set up as a function of a threshold value dif_swl, it then being the case that:
• αi,jn+1=α0i,jn+1*(1−Γi,jn+1),
• a weighting of the attenuation factor α0i,jn+1 is performed as a function of the movement measure Γi,jn+1,
• the recursive filtering of unmoved pixels with an attenuation factor α0i,jn+1 approaching α0max is performed, and
• the recursive filtering of moved pixels with an attenuation factor α0i,jn+1 approaching α0min is performed.