KeCoDe_degradation

PURPOSE ^

KeCoDe_degradation degrades the input image according to some

SYNOPSIS ^

function [degraded,PSF] = KeCoDe_degradation(Im,type,varargin);

DESCRIPTION ^

  KeCoDe_degradation degrades the input image according to some
  degradation model. The available degradation models include:

     * 'wt_gn'      : White Gaussian Noise
     * 'col_gn'     : Colored Gaussian Noise
     * 'blur_wt_gn' : Blur plus white Gaussian Noise
     * 'blur_col_gn': Blur plus colored Gaussian Noise
     * 'jpeg'       : JPEG degradation due to compression
     * 'salt_pepper': Salt and Pepper Noise.

  The number of input arguments depends on the degradation model
  They have to be included in the described order:

     * 'wt_gn'      : White Gaussian Noise
                      Needs (1) the variance of the noise
     * 'col_gn'     : Colored Gaussian Noise
                      Needs (1) the frequency band limits of the noise
                                [f_min f_max] expresed in normalized units
                                (in the range [0,1], where 1 is the
                                Nyquist frequency).
                            (2) The variance of the noise

     * 'blur_wt_gn' : Blur plus white Gaussian Noise
                      Needs (1) the cut-off frequency of the blurring
                                operator (in normalized units).
                            (2) the minimum value of the blurring operator
                                (in the range [0,1]).
                                In order to preserve some information of
                                high frequency components, the user can
                                introduce a minimum value for the amplitude
                                of the blurring filter (maximum attenuation
                                factor).
                                Value 0 means that the Gaussian low-pass
                                operator will not be modified.
                            (3) The variance of the noise.

     * 'blur_col_gn': Blur plus colored Gaussian Noise
                      Needs (1) the frequency band limits of the noise
                                [f_min f_max] expresed in normalized units
                                (in the range [0,1], where 1 is the
                                Nyquist frequency).
                            (2) the cut-off frequency of the blurring
                                operator (in normalized units).
                            (3) the minimum value of the blurring operator
                                (in the range [0,1]).
                                In order to preserve some information of
                                high frequency components, the user can
                                introduce a minimum value for the amplitude
                                of the blurring filter (maximum attenuation
                                factor).
                                Value 0 means that the Gaussian low-pass
                                operator will not be modified.
                            (4) The variance of the noise.

     * 'jpeg'       : JPEG degradation due to compression
                      Needs (1) the quality parameter (see imwrite)

     * 'salt_pepper': Salt and Pepper Noise.
                      Needs (1) noise density (see imnoise)

   OUTPUT:
   Id : degraded image
   PSF: the Point Spread Function of the blurring (this can be just
        a delta function if there is no blurring).

   SYNTAX:

   [degraded,PSF] = KeCoDe_degradation(Im,'Degradat_Model',Param1,Param2,...);

CROSS-REFERENCE INFORMATION ^

This function calls: This function is called by:

SOURCE CODE ^

0001 %
0002 %  KeCoDe_degradation degrades the input image according to some
0003 %  degradation model. The available degradation models include:
0004 %
0005 %     * 'wt_gn'      : White Gaussian Noise
0006 %     * 'col_gn'     : Colored Gaussian Noise
0007 %     * 'blur_wt_gn' : Blur plus white Gaussian Noise
0008 %     * 'blur_col_gn': Blur plus colored Gaussian Noise
0009 %     * 'jpeg'       : JPEG degradation due to compression
0010 %     * 'salt_pepper': Salt and Pepper Noise.
0011 %
0012 %  The number of input arguments depends on the degradation model
0013 %  They have to be included in the described order:
0014 %
0015 %     * 'wt_gn'      : White Gaussian Noise
0016 %                      Needs (1) the variance of the noise
0017 %     * 'col_gn'     : Colored Gaussian Noise
0018 %                      Needs (1) the frequency band limits of the noise
0019 %                                [f_min f_max] expresed in normalized units
0020 %                                (in the range [0,1], where 1 is the
0021 %                                Nyquist frequency).
0022 %                            (2) The variance of the noise
0023 %
0024 %     * 'blur_wt_gn' : Blur plus white Gaussian Noise
0025 %                      Needs (1) the cut-off frequency of the blurring
0026 %                                operator (in normalized units).
0027 %                            (2) the minimum value of the blurring operator
0028 %                                (in the range [0,1]).
0029 %                                In order to preserve some information of
0030 %                                high frequency components, the user can
0031 %                                introduce a minimum value for the amplitude
0032 %                                of the blurring filter (maximum attenuation
0033 %                                factor).
0034 %                                Value 0 means that the Gaussian low-pass
0035 %                                operator will not be modified.
0036 %                            (3) The variance of the noise.
0037 %
0038 %     * 'blur_col_gn': Blur plus colored Gaussian Noise
0039 %                      Needs (1) the frequency band limits of the noise
0040 %                                [f_min f_max] expresed in normalized units
0041 %                                (in the range [0,1], where 1 is the
0042 %                                Nyquist frequency).
0043 %                            (2) the cut-off frequency of the blurring
0044 %                                operator (in normalized units).
0045 %                            (3) the minimum value of the blurring operator
0046 %                                (in the range [0,1]).
0047 %                                In order to preserve some information of
0048 %                                high frequency components, the user can
0049 %                                introduce a minimum value for the amplitude
0050 %                                of the blurring filter (maximum attenuation
0051 %                                factor).
0052 %                                Value 0 means that the Gaussian low-pass
0053 %                                operator will not be modified.
0054 %                            (4) The variance of the noise.
0055 %
0056 %     * 'jpeg'       : JPEG degradation due to compression
0057 %                      Needs (1) the quality parameter (see imwrite)
0058 %
0059 %     * 'salt_pepper': Salt and Pepper Noise.
0060 %                      Needs (1) noise density (see imnoise)
0061 %
0062 %   OUTPUT:
0063 %   Id : degraded image
0064 %   PSF: the Point Spread Function of the blurring (this can be just
0065 %        a delta function if there is no blurring).
0066 %
0067 %   SYNTAX:
0068 %
0069 %   [degraded,PSF] = KeCoDe_degradation(Im,'Degradat_Model',Param1,Param2,...);
0070 %
0071 
0072 function [degraded,PSF] = KeCoDe_degradation(Im,type,varargin);
0073 
0074 warning('off','MATLAB:dispatcher:InexactMatch')
0075 if (strcmpi(type,'wt_gn'))
0076 
0077     fc = 3;
0078     Val_min = 0.001;
0079 
0080     lim_frec = [0 3];
0081 
0082     variance = varargin{1};
0083 
0084     [degraded,PSF,H,noise,Blurred]=degrade(Im,fc,Val_min,lim_frec,variance);
0085 
0086 elseif  (strcmpi(type,'col_gn'))
0087 
0088     fc = 3;
0089     Val_min = 0.001;
0090 
0091     lim_frec = varargin{1};
0092 
0093     variance = varargin{2};
0094 
0095     [degraded,PSF,H,noise,Blurred]=degrade(Im,fc,Val_min,lim_frec,variance);
0096 
0097 elseif  (strcmpi(type,'blur_wt_gn'))
0098 
0099     fc = varargin{1};
0100     Val_min = varargin{2};
0101 
0102     lim_frec = [0 3];
0103 
0104     variance = varargin{3};
0105 
0106     [degraded,PSF,H,noise,Blurred]=degrade(Im,fc,Val_min,lim_frec,variance);
0107 
0108 elseif  (strcmpi(type,'blur_col_gn'))
0109 
0110     fc = varargin{1};
0111     Val_min = varargin{2};
0112 
0113     lim_frec = varargin{3};
0114 
0115     variance = varargin{4};
0116 
0117     [degraded,PSF,H,noise,Blurred]=degrade(Im,fc,Val_min,lim_frec,variance);
0118 
0119 elseif  (strcmpi(type,'jpeg'))
0120 
0121     you_are_here=pwd;
0122     output_folder=which('KeCoDe_coder_tools_demo_achrom');
0123     output_folder=output_folder(1:end-32);
0124     cd(output_folder);
0125     cd('tmp');
0126 
0127     PSF = zeros(11);
0128     PSF(6,6)=1;
0129 
0130     imwrite(Im/255,'tmp.jpg','Quality',varargin{1});
0131     degraded = imread('tmp.jpg');
0132 
0133     cd(you_are_here);
0134     warning off
0135 
0136 elseif  (strcmpi(type,'salt_pepper'))
0137 
0138     PSF = zeros(11);
0139     PSF(6,6)=1;
0140     degraded = imnoise(Im/255,'salt & pepper', varargin{1});
0141     degraded = degraded*255;
0142 else
0143     error('The requested degradation is not implemented');
0144 end
0145 
0146 degraded = double(degraded);

Generated on Fri 07-Mar-2014 13:28:33 by m2html © 2005