PERF: Remove allocation on image casting

Summary

A lot of non necessary allocations are done almost every time an image is read.

This cost can be removed.

Rationale

No need to pay for computations that can be avoided.

On simple applications, a x2 speed up is easily observed. It was the case with https://gitlab.orfeo-toolbox.org/s1-tiling/normlim_sigma0 SARComputeLocalIncidenceAngle that reads 2 images, and computes for each pixel 1 cross product, 2 norms, one arcsin.

This MR relates to #2065 (closed).

Implementation Details

Classes and files

• A new class to iterate over pixel components is provided: PixelComponentIterator. For instance this iterator is able to convert a FixedArray<std::complex<float>> into a VariableLengthArray<double>
• ClampFilter architecture is changed to efficiently handle VariableLengthVector pixels: it's no longer an UnaryFunctorImageFilter (that calls result = f(in) internally) but a plain filter that calls f(result, in) on each pixel. This permits to avoid creating (and destroying) a new VLV for each pixel, and thus to avoid dynamic allocations. The filter has also been simplified and optimized.
• ConvertTypeFunctor has been adapted for the new interface of ClampFilter. Internally, it uses the new PixelComponentIterator to copy components from the input pixel to the output pixel. Along the way, it has been simplified and optimized.
• CastImage application has been simplified. Instead of converting any image into a VectorImage<double> of the same number of components/bands, and then into the output image, it directly converts the input image into the output image -- thanks to ClampFilter.
• WrapperInputImageParameter and WrapperOutputImageParameter have been simplified

Also a few new helpers functions and classes have been provided:

• like a C++14 backport of C++17 std::clamp
• a otb::is_array traits that recognizes itk::VLV and classes that inherits from itk::FixedArray
• GetNumberOfComponents() that works around shortcomings of itk::DefaultConvertPixelTraits<complex<>>::GetNumberOfComponents(pix) (which doesn't support std::complex and that copies VLV (see https://github.com/InsightSoftwareConsortium/ITK/issues/2312))
• a port of gsl::not_null into OTB.

Tests

Old tests for ClampFilter (bfTvClampImageFilterConversionTest) still runs.

A new test for PixelComponentIterator has been added.

Copyright

The copyright owner is CSGroup FRANCE and has signed the ORFEO ToolBox Contributor License Agreement.

---

Check before merging:

• All discussions are resolved
• At least 2 votes from core developers, no vote.
• The feature branch is (reasonably) up-to-date with the base branch
• Dashboard is green
• Copyright owner has signed the ORFEO ToolBox Contributor License Agreement
• Optionally, run git diff develop... -U0 --no-color | clang-format-diff.py -p1 -i on latest changes and commit

added 1 commit

• 82ff7f71 - COMP: Try to please clang 6

Compare with previous version

added 1 commit

• 0bdeb54a - COMP: Try to please old clang 6

Compare with previous version

added 2 commits

• a67dd0e0 - COMP: Add missing include
• 22c27256 - BUG: Restore cache of casted images

Compare with previous version

mentioned in merge request !799

added 1 commit

• c7bfc855 - BUG: Fix narrowing clampling

Compare with previous version

mentioned in issue #2155 (closed)

Regarding the tests failing on Windows with vc14, I'm waiting to see how !799 evolves: hopefully the current MR should compile with a version of vc compliant with C++17.

@lhermitte : Ok, I've commented !799 before having a look at this MR. I understand the rationale of this MR and this sounds good to implement the corrections you propose, but maybe we have to discuss of the possible side effects...

@ytanguy Note that the code that doesn't compile on Windows is valid C++14 code (g++ and clang++ would have rejected it otherwise). It's just that vc14 is not fully compliant with C++14...

According to this table, we need to use at leastVisual Studio 2017 15.0 (1910, MSVC++14.1 ) to have full c++ 14 support. If msvc 14 does not fully support c++ 14, I think it makes sense to drop support for this compiler.

added 179 commits

• c7bfc855...bc0cb8dd - 178 commits from branch orfeotoolbox:develop
• 80bd885e - Merge branch 'develop' of https://gitlab.orfeo-toolbox.org/orfeotoolbox/otb...

Compare with previous version

merged

mentioned in commit 807114a7

mentioned in merge request !827 (merged)