Using fscanf when some elements are not floating point

Hi there

I am trying to use fscanf to read a measurement text file. The file is a matrix of (480x640) floating point numbers. Normally this would be fine however some elements are strings (in this case 'Bad') - is there a way of reading the file in the same way but replacing 'Bad' with NaN, or a better way of doing this?

Currently I have used the following, which reads to the point of the first 'Bad'.

B=fscanf(fid,'%f ',[480 640])

Many thanks in advance

Jack

Answer 1

[Cedric]

One way to do it is to start with string replacement.. to illustrate, say that the data file contains

 4   5   6   7   8   9
 1   2   Bad 4   5   6
 Bad 7   8   9   10  11
 0   Bad Bad Bad 4   5

then:

 >> buffer = fileread('myData.txt') ;
 >> buffer = strrep(buffer, 'Bad', 'NaN') ;    % or regexprep().
 >> sscanf(buffer, '%f', [6 4]).'
 ans =
     4     5     6     7     8     9
     1     2   NaN     4     5     6
   NaN     7     8     9    10    11
     0   NaN   NaN   NaN     4     5

[Jan]

@jnaumann: Just to repeat the obvious solution: Replace the not recognized "Bad" by the recognized "NaN" in the file. Then the reading is straight forward. Sometimes simplicity is really nice. +1