Field
     Image processing, Method for information recording and read/write method thereof,
     Measuring method, measurement results display

Keywords
     Image restoration, a Kalman filter,blur,noise, two-dimensional block Kalman filter,
     auto regressive, colored driving source, High-speed operation, 
     State space model without using the AR system

Advantages
      This patent proposes a robust image restoration method using two-dimensional block Kalman
     filter with colored driving source. This method aims to achieve high quality image restoration
     for blur and noise disturbance from the canonical state space models with (i) a state equation
     composed of the original image, and (ii) an observation equation composed of the original
     image, blur, and noise. The remarkable feature of the proposed method is realization of high
     performance image restoration without sacrificing original image despite simple image
     restoration using only Kalman filter. We show the effectiveness of the proposed method, using
     numerical results and subjective evaluation results.
      Current Technology is a preventing degradation method. This technology is restoration of
     degraded images, using the improved Kalman Filter technique. 
     Since preconditions existed in Kalman filter theory, it had taken time to operate a Kalman
     filter algorithm.　 The processing of this method became a quick operation. And a restoration
     is more clear.
    
     Picture size       Kalman Filter      This Method
        ２５６ｘ２５６　　　   20.8sec.             5.3sec.
        ５１２ｘ５１２　　　   83.1sec.　　　　  　21.1sec.

Abstract
A simple and practical image restoring device capable of improving the performance of image restoration. An image restoring device (100) is provided with a first restoration processing unit (160a) and estimates original image information from the degraded image information into which information unnecessary for the original image information is incorporated. A correlation calculating section (164) calculates the correlation value of the estimation error of when the state amount of a system at time n+1 including the original image information is estimated by information up to time n or the time n+1 with respect to the degraded image information of only the time n. A weighting factor calculating section (166) calculates a weighting factor to define the relationship of the optimum estimated value of the state amount at the time n+1 by the information up to the time n+1, the optimum estimated value of the state amount at the time n+1 by the information up to the time n, and an estimation error of an observed amount including the degraded image information by using the correlation value with respect to the degraded image information of only the time n. An optimum estimated value calculating section (168) calculates the optimum estimated value of the state amount at the time n+1 by the information up to the time n+1 by using the weighting factor with respect to the degraded image information of only the time n.

         
Inventor
     Dr.　Nari Tanabe　(Tokyo University of Science)


Patent Publication No
     (APPLICATION NUMBER, FILING DATE)
     WO2009003780 (Aug. 6, 2009), EP09804750.9（Aug. 6. 2009), US13/057059 (Aug. 6, 20092009),
     JP2010-523766 (Aug. 6, 2009)
     (PUBLICATION NUMBER, PUBLICATION DATE)
     WO2010/016263 (Feb. 11, 2010),  EP2302584 （Mar. 30, 2011）


Internal file number
     LSP:77
     G2009-004 (T2008-034), T2010-065

To access the rest of the information on this page,
Log In or sign up for a free B-Bridge account: