Changes to DIPimage
- A new option 'KeepDataType' in dipsetpref causes arithmetic operations to produce an output image of the same type as the input. This option should be set to 'on' when memory is limited and images are large. Note that overflow can occur and will not be detected!
- dipgetpref now returns boolean preference values as a boolean rather than a string 'on'/'off', for ease of use.
- Fixed bug in mdhistogram where using the bins_format input argument only worked properly for 1D.
- Output images from DIPlib functions were copied in the DIP-MEX interface. We're not sure since which version of MATLAB this happened, but it won't happen any more.
Changes to DIPlib
(These might or might not propagate to DIPimage)
- Fixed bug with 'zero order hold' and 'nearest neighbour' interpolation methods: the shift was applied after rather than before the zoom (as is the case with the other methods).
- Fixed dip_Mean() and the like so that pixels where the mask has a value of 0 are not used in the computation; this lead to NaN values propagating to the output even if they were "masked out". Also fixed bug in some of the functions that use the mask as weights, but tested for the mask to be a binary image; it is now possible for example to compute weighted means.