Description of slgetinterpkernel

Home > sltoolbox > interp > slgetinterpkernel.m

slgetinterpkernel

PURPOSE

SLGETINTERPKERNEL Gets the interpolation kernel function

SYNOPSIS

function [f, r] = slgetinterpkernel(kername)

DESCRIPTION

SLGETINTERPKERNEL Gets the interpolation kernel function

 $ Syntax $
   - [f, r] = slgetinterpkernel(kername)
   
 $ Arguments $
   - kername:      The name of the interpolation kernel
   - f:            The function handle to the kernel
   - r:            The effective radius of the kernel

 $ Description $
   - [f, r] = slgetinterpkernel(kername) gets the function handle to 
     an interpolation kernel and the corresponding effective radius. 
     The supported kernel include:
       - 'nearest':         The nearest neighbor interpolation
       - 'linear':          The linear interpolation
       - 'cubic':           The cubic interpolation
     For generality, the kername can also be a cell array as
     {f, r}. Then the function directly extracts them to output.
     Here are the formulas for the kernels:
       - 'nearest':        f(x) = 1, when |x| <= 0.5
                                  0, when |x| > 0.5 
       - 'linear':         f(x) = 1 - |x|, when |x| <= 1
                                  0,       when |x| > 1
       - 'cubic':          f(x) = 1 - 2|x|^2 + |x|^3,        when |x| <= 1
                                  4 - 8|x| + 5|x|^2 - |x|^3, when 1 < |x| <= 2
                                  0,                         when |x| > 2

 $ Remarks $
   - All the kernel functions are vectorized, so they all support
     both scalar input and array input of any dimension.
   - The kernel functions are legal only within the effective radius,
     in the outside region, the produced values are undefined (not
     necessary zero). Such a design is for efficiency, so it is the
     invoker's responsibility to guarantee the input is in valid 
     range.

 $ History $
   - Created by Dahua Lin, on Sep 2nd, 2006

CROSS-REFERENCE INFORMATION

This function calls:

This function is called by:

• slimginterp SLIMGINTERP Performs image based interpolation

SUBFUNCTIONS

• function y = cubic_interpolant(x)

SOURCE CODE

0001 function [f, r] = slgetinterpkernel(kername)
0002 %SLGETINTERPKERNEL Gets the interpolation kernel function
0003 %
0004 % $ Syntax $
0005 %   - [f, r] = slgetinterpkernel(kername)
0006 %
0007 % $ Arguments $
0008 %   - kername:      The name of the interpolation kernel
0009 %   - f:            The function handle to the kernel
0010 %   - r:            The effective radius of the kernel
0011 %
0012 % $ Description $
0013 %   - [f, r] = slgetinterpkernel(kername) gets the function handle to
0014 %     an interpolation kernel and the corresponding effective radius.
0015 %     The supported kernel include:
0016 %       - 'nearest':         The nearest neighbor interpolation
0017 %       - 'linear':          The linear interpolation
0018 %       - 'cubic':           The cubic interpolation
0019 %     For generality, the kername can also be a cell array as
0020 %     {f, r}. Then the function directly extracts them to output.
0021 %     Here are the formulas for the kernels:
0022 %       - 'nearest':        f(x) = 1, when |x| <= 0.5
0023 %                                  0, when |x| > 0.5
0024 %       - 'linear':         f(x) = 1 - |x|, when |x| <= 1
0025 %                                  0,       when |x| > 1
0026 %       - 'cubic':          f(x) = 1 - 2|x|^2 + |x|^3,        when |x| <= 1
0027 %                                  4 - 8|x| + 5|x|^2 - |x|^3, when 1 < |x| <= 2
0028 %                                  0,                         when |x| > 2
0029 %
0030 % $ Remarks $
0031 %   - All the kernel functions are vectorized, so they all support
0032 %     both scalar input and array input of any dimension.
0033 %   - The kernel functions are legal only within the effective radius,
0034 %     in the outside region, the produced values are undefined (not
0035 %     necessary zero). Such a design is for efficiency, so it is the
0036 %     invoker's responsibility to guarantee the input is in valid
0037 %     range.
0038 %
0039 % $ History $
0040 %   - Created by Dahua Lin, on Sep 2nd, 2006
0041 %
0042 
0043 %% Main skeleton
0044 
0045 if ischar(kername)
0046     switch kername
0047         case 'nearest'
0048             f = @(x) 1;
0049             r = 0.5;
0050         case 'linear'
0051             f = @(x) 1 - abs(x);
0052             r = 1;
0053         case 'cubic'
0054             f = @cubic_interpolant;
0055             r = 2;
0056         otherwise
0057             error('sltoolbox:invalidarg', ...
0058                 'The interpolation kernel name is unsupported: %s', kername);
0059     end
0060 elseif iscell(kername)
0061     f = kername{1};
0062     r = kername{2};
0063     if ~isa(f, 'funtion_handle')
0064         error('sltoolbox:invalidarg', ...
0065             'The interpolation kernel is invalid');
0066     end
0067 else
0068     error('sltoolbox:invalidarg', ...
0069             'The interpolation kernel is invalid');
0070 end
0071         
0072 
0073 
0074 %% Core functions
0075 
0076 function y = cubic_interpolant(x)
0077 
0078 x = abs(x);
0079 p1 = 1 - x .* x .* (2 - x);
0080 p2 = 4 + x .* (-8 + x .* (5 - x));
0081 b1 = x < 1;
0082 b2 = true - b1;
0083 y = p1 .* b1 + p2 .* b2;
0084 
0085 
0086         
0087         
0088         
0089         
0090

---

Generated on Wed 20-Sep-2006 12:43:11 by m2html © 2003