ABSTRACT
PSF values, edge detection and morphological dilation and region filling technology, rebuild the restored image, improved the quality of the recovery and reduced the phenomenon of ring [3].
2 COMMON RECOVERY MODELS
2.1 Weiner filtering and constrained least squares filtering restoration model
The Weiner filtering was first proposed by N. Wiener in 1942. The method is based on the view that image and noise are random processes. The goal is to find an estimate f
∧ of the pollution image
f to make a minimum mean square error between them. Error metric formula: e E f f2 2[( ) ]
∧ .
Where E[ ] is the expectation value of the variable. Assuming noise and image are uncorrelated, one of them has a zero mean. Estimated grayscale is a linear function of the deteriorated image grayscale. Under these conditions, the minimum errors function in the frequency domain is by the following calculation formulas:
