matlab画图nan,在Matlab中过滤包含NaN的图像？

我最终使用的技术是Matlab的File Exchange中的功能

nanconv.m.它完全符合我的要求&＃xff1a;它以一种忽略NaN的方式运行过滤器,就像Matlab的内置函数nanmean那样.这很难从功能的文档中解读,这有点神秘.

这是我如何使用它&＃xff1a;

filtWidth &＃61; 7;

filtSigma &＃61; 5;

imageFilter&＃61;fspecial(&＃39;gaussian&＃39;,filtWidth,filtSigma);

dataFiltered &＃61; nanconv(data,imageFilter, &＃39;nanout&＃39;);

我正在粘贴下面的nanconv函数(它由the BSD license覆盖).当我有机会的时候,我会张贴图片等,只是想发布我最后做的事情,对于那些好奇我做了什么的人.

与其他答案的比较

使用gnovice’s solution,结果看起来非常直观,但是边缘有一些定量的亮点是一个值得关注的问题.在实践中,超出边缘的图像外推导致我的数据边缘处的许多虚假的高值.

使用krisdestruction’s suggestion用原始数据替换丢失的位,看起来也相当不错(特别是对于非常小的滤波器),但是(按设计)你最终会在边缘处得到未经滤波的数据,这对我的应用来说是一个问题.

nanconv

function c &＃61; nanconv(a, k, varargin)

% NANCONV Convolution in 1D or 2D ignoring NaNs.

% C &＃61; NANCONV(A, K) convolves A and K, correcting for any NaN values

% in the input vector A. The result is the same size as A (as though you

% called &＃39;conv&＃39; or &＃39;conv2&＃39; with the &＃39;same&＃39; shape).

%

% C &＃61; NANCONV(A, K, &＃39;param1&＃39;, &＃39;param2&＃39;, ...) specifies one or more of the following:

% &＃39;edge&＃39; - Apply edge correction to the output.

% &＃39;noedge&＃39; - Do not apply edge correction to the output (default).

% &＃39;nanout&＃39; - The result C should have NaNs in the same places as A.

% &＃39;nonanout&＃39; - The result C should have ignored NaNs removed (default).

% Even with this option, C will have NaN values where the

% number of consecutive NaNs is too large to ignore.

% &＃39;2d&＃39; - Treat the input vectors as 2D matrices (default).

% &＃39;1d&＃39; - Treat the input vectors as 1D vectors.

% This option only matters if &＃39;a&＃39; or &＃39;k&＃39; is a row vector,

% and the other is a column vector. Otherwise, this

% option has no effect.

%

% NANCONV works by running &＃39;conv2&＃39; either two or three times. The first

% time is run on the original input signals A and K, except all the

% NaN values in A are replaced with zeros. The &＃39;same&＃39; input argument is

% used so the output is the same size as A. The second convolution is

% done between a matrix the same size as A, except with zeros wherever

% there is a NaN value in A, and ones everywhere else. The output from

% the first convolution is normalized by the output from the second

% convolution. This corrects for missing (NaN) values in A, but it has

% the side effect of correcting for edge effects due to the assumption of

% zero padding during convolution. When the optional &＃39;noedge&＃39; parameter

% is included, the convolution is run a third time, this time on a matrix

% of all ones the same size as A. The output from this third convolution

% is used to restore the edge effects. The &＃39;noedge&＃39; parameter is enabled

% by default so that the output from &＃39;nanconv&＃39; is identical to the output

% from &＃39;conv2&＃39; when the input argument A has no NaN values.

%

% See also conv, conv2

%

% AUTHOR: Benjamin Kraus (bkraus&＃64;bu.edu, ben&＃64;benkraus.com)

% Copyright (c) 2013, Benjamin Kraus

% $Id: nanconv.m 4861 2013-05-27 03:16:22Z bkraus $

% Process input arguments

for arg &＃61; 1:nargin-2

switch lower(varargin{arg})

case &＃39;edge&＃39;; edge &＃61; true; % Apply edge correction

case &＃39;noedge&＃39;; edge &＃61; false; % Do not apply edge correction

case {&＃39;same&＃39;,&＃39;full&＃39;,&＃39;valid&＃39;}; shape &＃61; varargin{arg}; % Specify shape

case &＃39;nanout&＃39;; nanout &＃61; true; % Include original NaNs in the output.

case &＃39;nonanout&＃39;; nanout &＃61; false; % Do not include NaNs in the output.

case {&＃39;2d&＃39;,&＃39;is2d&＃39;}; is1D &＃61; false; % Treat the input as 2D

case {&＃39;1d&＃39;,&＃39;is1d&＃39;}; is1D &＃61; true; % Treat the input as 1D

end

end

% Apply default options when necessary.

if(exist(&＃39;edge&＃39;,&＃39;var&＃39;)~&＃61;1); edge &＃61; false; end

if(exist(&＃39;nanout&＃39;,&＃39;var&＃39;)~&＃61;1); nanout &＃61; false; end

if(exist(&＃39;is1D&＃39;,&＃39;var&＃39;)~&＃61;1); is1D &＃61; false; end

if(exist(&＃39;shape&＃39;,&＃39;var&＃39;)~&＃61;1); shape &＃61; &＃39;same&＃39;;

elseif(~strcmp(shape,&＃39;same&＃39;))

error([mfilename &＃39;:NotImplemented&＃39;],&＃39;Shape &＃39;&＃39;%s&＃39;&＃39; not implemented&＃39;,shape);

end

% Get the size of &＃39;a&＃39; for use later.

sza &＃61; size(a);

% If 1D, then convert them both to columns.

% This modification only matters if &＃39;a&＃39; or &＃39;k&＃39; is a row vector, and the

% other is a column vector. Otherwise, this argument has no effect.

if(is1D);

if(~isvector(a) || ~isvector(k))

error(&＃39;MATLAB:conv:AorBNotVector&＃39;,&＃39;A and B must be vectors.&＃39;);

end

a &＃61; a(:); k &＃61; k(:);

end

% Flat function for comparison.

o &＃61; ones(size(a));

% Flat function with NaNs for comparison.

on &＃61; ones(size(a));

% Find all the NaNs in the input.

n &＃61; isnan(a);

% Replace NaNs with zero, both in &＃39;a&＃39; and &＃39;on&＃39;.

a(n) &＃61; 0;

on(n) &＃61; 0;

% Check that the filter does not have NaNs.

if(any(isnan(k)));

error([mfilename &＃39;:NaNinFilter&＃39;],&＃39;Filter (k) contains NaN values.&＃39;);

end

% Calculate what a &＃39;flat&＃39; function looks like after convolution.

if(any(n(:)) || edge)

flat &＃61; conv2(on,k,shape);

else flat &＃61; o;

end

% The line above will automatically include a correction for edge effects,

% so remove that correction if the user does not want it.

if(any(n(:)) && ~edge); flat &＃61; flat./conv2(o,k,shape); end

% Do the actual convolution

c &＃61; conv2(a,k,shape)./flat;

% If requested, replace output values with NaNs corresponding to input.

if(nanout); c(n) &＃61; NaN; end

% If 1D, convert back to the original shape.

if(is1D && sza(1) &＃61;&＃61; 1); c &＃61; c.&＃39;; end

end