Signal processing apparatus and method using local length scales for deblurring

Abstract

A signal processing apparatus for deblurring a digital input signal (I(x i )) comprising one or more processors. The one or more processors are configured to: compute a plurality of local length scales (l k , l(x i ), λ) from at least one of a local signal resolution (FRC k ) and a local signal-to-noise ratio (SNR k ), and to compute each of the local length scales at a different location (I k (x i )) of the digital input signal, each of the different locations comprising at least one sample point (x i ) of the input signal; compute a baseline estimate (ƒ n (x i , l n ), ƒ(x i , l(x i ))) of the input signal based on the local length scales, the baseline estimate representing signal structures of the digital input signal that are larger than the local length scale; and compute a digital output signal (O(x i )) based on one of (a) the baseline estimate and (b) the digital input signal and the baseline estimate.

1 : A signal processing apparatus for deblurring a digital input signal (I(x i )), the signal processing apparatus comprising one or more processors configured to:
 compute a plurality of local length scales (l k , l(x i ), λ) from at least one of a local signal resolution (FRC k ) and a local signal-to-noise ratio (SNR), and to compute each of the local length scales at a different location (I k (x i )) of the digital input signal, each of the different locations comprising at least one sample point (x i ) of the input signal;   compute at least one baseline estimate (ƒ n (x i , l n ), ƒ(x i , l(x i ))) of the input signal based on the local length scales, the at least one baseline estimate representing signal structures of the digital input signal that are larger than the local length scale; and   compute a digital output signal (O(x i )) based on one of: (a) the at least one baseline estimate and (b) the digital input signal and the at least one baseline estimate.   
     
     
         2 : The signal processing apparatus according to  claim 1 , wherein the signal processing apparatus is configured to compute each of the local length scales (l k , l(x i ), λ) based on a Fourier Ring Correlation. 
     
     
         3 : The signal processing apparatus according to  claim 1 , wherein the signal processing apparatus is configured to compute at least one of the local length scales (l k , l(x i ), λ) based on a local signal-to-noise ratio (SNR(x i )). 
     
     
         4 : The signal processing apparatus according to  claim 1 , wherein the signal processing apparatus is further configured to compute each of the local length scales (l(x i )λ) in a location (I k (x i )) comprising a plurality of sample points (x i ). 
     
     
         5 : The signal processing apparatus according to  claim 4 , wherein the signal processing apparatus is further configured to interpolate the local length scales (l k ) computed at the respective location (I k (x i )) that comprises the respective plurality of sample points (x i ) to compute the local length scales (l(x i ), λ) at each of the sample points (x i ). 
     
     
         6 : The signal processing apparatus according to  claim 1 , wherein the signal processing apparatus is configured to:
 compute a local length scale (l(x i ), λ) at each sample point (x i ) of the digital input signal (I(x i )); and   compute a baseline estimate (ƒ(x i , l(x i ))) based on the local length scale (l(x i )) at each of the sample points of the digital input image.   
     
     
         7 : The signal processing apparatus according to  claim 1 , wherein the signal processing apparatus is configured to compute a plurality of baseline estimates (ƒ n (x i , l n )) of the digital input signal (I(x i )), each of the baseline estimates being based on a different constant local length scale (l n , λ). 
     
     
         8 : The signal processing apparatus according to  claim 5 , wherein the signal processing apparatus is configured to compute a further baseline estimate (ƒ(x i , l(x i ))) from the plurality of baseline estimates (ƒ n (x i , l n )) that are based on the different constant local length scales (l n , λ). 
     
     
         9 : The signal processing apparatus according to  claim 7 , wherein the signal processing apparatus is configured to;
 remove each of the plurality of baseline estimates (ƒ n (x i , l n )) that are based on the different constant local length scales (l n , λ) separately from the digital input signal (I(x i )) to obtain a plurality of intermediate output signals (J n (x i )); and   obtain the digital output signal (O(x i )) based on a combination of the plurality of intermediate output signals.   
     
     
         10 : The signal processing apparatus according to  claim 9 , wherein the signal processing apparatus is configured to interpolate at least two intermediate output signals (J n (x i )) at each sample point (x i ) to compute the digital output signal (O(x i )) at a matching sample point (x i ). 
     
     
         11 : The signal processing apparatus according to  claim 1 , wherein the signal processing apparatus is configured to compute the baseline estimate using a regularization in which the local length scale (l n , l(x i ), λ) is represented. 
     
     
         12 : The signal processing apparatus according to  claim 1 , wherein the local length scale is contained in a least-square minimization criterion. 
     
     
         13 : The signal processing apparatus according to  claim 12 , wherein the local length scale is contained in a penalty term of the least-square minimization criterion, the penalty term comprising a derivative of the baseline estimate. 
     
     
         14 : A signal processing method for deblurring a digital input signal (I(x i )), the signal processing method comprising:
 computing a plurality of local length scales (l k , l(x i ), λ) from at least one of a local signal resolution (FRC k ) and a local signal-to-noise ratio (SNR k ), each of the local length scales being computed at a different location (I k (x i )) of the digital input signal, each different location comprising at least one sample point (x i ) of the input signal;   computing at least one baseline estimate (ƒ n (x i , l n ), ƒ(x i , l(x i ))) of the input signal based on the local length scales, the at least one baseline estimate representing signal structures of the digital input signal that are larger than the local length scale; and   computing a digital output signal (O(x i )) based on one of (a) the at least one baseline estimate and (b) the digital input signal and the at least one baseline estimate.   
     
     
         15 . (canceled) 
     
     
         16 : A non-transitory computer readable medium storing a computer program causing a computer to execute the signal processing method according to  claim 14 . 
     
     
         17 : A neural network device trained by a plurality of digital input signals (I(x i )) and at least one of a plurality of digital output signals (J(x i )) created from the digital input signals by the method of  claim 14   
     
     
         18 : A digital output signal (O(x i )) that is the result of the signal processing method according to  claim 14 .