From eb733634a57e6885bef28a4453c85b412f76194c Mon Sep 17 00:00:00 2001
From: Christophe Palmann <christophe.palmann@c-s.fr>
Date: Fri, 4 Sep 2015 16:09:06 +0200
Subject: [PATCH] ENH: cleaning
---
...varianceToReciprocalCoherencyImageFilter.h | 53 ++++----------
...oReciprocalCircularCovarianceImageFilter.h | 73 +++++--------------
...eciprocalCircularCovarianceMatrixFunctor.h | 6 --
...clairToReciprocalCovarianceMatrixFunctor.h | 13 +---
4 files changed, 36 insertions(+), 109 deletions(-)
diff --git a/Modules/Filtering/Polarimetry/include/otbReciprocalCovarianceToReciprocalCoherencyImageFilter.h b/Modules/Filtering/Polarimetry/include/otbReciprocalCovarianceToReciprocalCoherencyImageFilter.h
index 698b6f21b7..f34b83c503 100644
--- a/Modules/Filtering/Polarimetry/include/otbReciprocalCovarianceToReciprocalCoherencyImageFilter.h
+++ b/Modules/Filtering/Polarimetry/include/otbReciprocalCovarianceToReciprocalCoherencyImageFilter.h
@@ -30,12 +30,12 @@ namespace Functor {
* \brief Evaluate the Coherency matrix from the Covariance image
*
* Output value are:
- * - channel #0 : \f$ 0.5 * (S_{hh}+S_{vv}.(S_{hh}+S_{vv})^{*} \f$
- * - channel #1 : \f$ 0.5 * (S_{hh}+S_{vv}.(S_{hh}-S_{vv})^{*} \f$
- * - channel #2 : \f$ (S_{hh}+S_{vv}.(S_{hv})^{*} \f$
- * - channel #3 : \f$ 0.5 * (S_{hh}-S_{vv}.(S_{hh}-S_{vv})^{*} \f$
- * - channel #4 : \f$ (S_{hh}-S_{vv}.(S_{hv})^{*} \f$
- * - channel #5 : \f$ 2.0*S_{hv}.S_{hv}^{*} \f$
+ * - channel #0 : \f$ 0.5 . (C33 + C13 + C13^{*} + C11) \f$
+ * - channel #1 : \f$ 0.5 . (-C33 - C13 + C13^{*} + C11) \f$
+ * - channel #2 : \f$ 0.5 . (\sqrt{2}.C12 + \sqrt{2}.C23^{*}) \f$
+ * - channel #3 : \f$ 0.5 . (C33 - C13 - C13^{*} + C11 \f$)
+ * - channel #4 : \f$ 0.5 . (\sqrt{2}.C12 - \sqrt{2}.C23^{*}) \f$
+ * - channel #5 : \f$ 0.5 . (2 . C22) \f$
*
* The output pixel has 6 channels : the diagonal and the upper element of the reciprocal matrix.
* Element are stored from left to right, line by line.
@@ -61,48 +61,23 @@ public:
TOutput result;
result.SetSize(m_NumberOfComponentsPerPixel);
- /* Using the convention
- * \f$ C_{11} = S_{hh}*S_{hh}^* \f$
- * \f$ C_{12} = S_{hh}*S_{hv}^* \f$
- * \f$ C_{13} = S_{hh}*S_{vv}^* \f$
- * \f$ C_{22} = S_{hv}*S_{hv}^* \f$
- * \f$ C_{23} = S_{hv}*S_{vv}^* \f$
- * \f$ C_{33} = S_{vv}*S_{vv}^* \f$
- */
- /*const ComplexType C11 = static_cast<ComplexType>(Covariance[0]);
+
+ const ComplexType C11 = static_cast<ComplexType>(Covariance[0]);
const ComplexType C12 = static_cast<ComplexType>(Covariance[1]);
const ComplexType C13 = static_cast<ComplexType>(Covariance[2]);
const ComplexType C22 = static_cast<ComplexType>(Covariance[3]);
const ComplexType C23 = static_cast<ComplexType>(Covariance[4]);
const ComplexType C33 = static_cast<ComplexType>(Covariance[5]);
-
- //const ComplexType C21 = vcl_conj(C12);
- const ComplexType C31 = vcl_conj(C13);
- const ComplexType C32 = vcl_conj(C23);
-
- result[0] = static_cast<OutputValueType>( 0.5*(C11 + C13 + C31 + C33) );
- result[1] = static_cast<OutputValueType>( 0.5*(C11 - C13 + C31 - C33) );
- result[2] = static_cast<OutputValueType>( C12 + C32 );
- result[3] = static_cast<OutputValueType>( 0.5*(C11 - C13 - C31 + C33) );
- result[4] = static_cast<OutputValueType>( C12 - C32 );
- result[5] = static_cast<OutputValueType>( 2.0 * C22 );*/
-
- const ComplexType A = static_cast<ComplexType>(Covariance[0]);
- const ComplexType B = static_cast<ComplexType>(Covariance[1]);
- const ComplexType C = static_cast<ComplexType>(Covariance[2]);
- const ComplexType E = static_cast<ComplexType>(Covariance[3]);
- const ComplexType F = static_cast<ComplexType>(Covariance[4]);
- const ComplexType I = static_cast<ComplexType>(Covariance[5]);
const ComplexType two = ComplexType(2.0, 0.0);
const ComplexType rootTwo = ComplexType(vcl_sqrt(2.0), 0.0);
- result[0] = static_cast<OutputValueType>( I + C + vcl_conj(C) + A );
- result[1] = static_cast<OutputValueType>( -I - C + vcl_conj(C) + A );
- result[2] = static_cast<OutputValueType>( rootTwo*B + rootTwo*vcl_conj(F) );
- result[3] = static_cast<OutputValueType>( I - C - vcl_conj(C) + A );
- result[4] = static_cast<OutputValueType>( rootTwo*B - rootTwo*vcl_conj(F) );
- result[5] = static_cast<OutputValueType>( two * E );
+ result[0] = static_cast<OutputValueType>( C33 + C13 + vcl_conj(C13) + C11 );
+ result[1] = static_cast<OutputValueType>( -C33 - C13 + vcl_conj(C13) + C11 );
+ result[2] = static_cast<OutputValueType>( rootTwo*C12 + rootTwo*vcl_conj(C23) );
+ result[3] = static_cast<OutputValueType>( C33 - C13 - vcl_conj(C13) + C11 );
+ result[4] = static_cast<OutputValueType>( rootTwo*C12 - rootTwo*vcl_conj(C23) );
+ result[5] = static_cast<OutputValueType>( two * C22 );
result /= 2.0;
diff --git a/Modules/Filtering/Polarimetry/include/otbReciprocalLinearCovarianceToReciprocalCircularCovarianceImageFilter.h b/Modules/Filtering/Polarimetry/include/otbReciprocalLinearCovarianceToReciprocalCircularCovarianceImageFilter.h
index 2650559f81..8a8caa7b31 100644
--- a/Modules/Filtering/Polarimetry/include/otbReciprocalLinearCovarianceToReciprocalCircularCovarianceImageFilter.h
+++ b/Modules/Filtering/Polarimetry/include/otbReciprocalLinearCovarianceToReciprocalCircularCovarianceImageFilter.h
@@ -32,23 +32,14 @@ namespace Functor {
* Extract from Antennas for radar and communications Harold Mott p 317.
*
* Output value are:
- * - channel #0 : \f$ 0.25 * (C_{1}+C_{3}+4*C_{2}-2*C_{6}-4*C_{5}-4*C_{9}) \f$
- * - channel #1 : \f$ 0.25 * (C_{1}-C_{3}+2*C_{5}+2*C_{9} - 0.5\mathcal{i}.(C_{4}+C_{7}+C_{8}) \f$
- * - channel #2 : \f$ 0.25 * (C_{1}+C_{3}-4*C_{2}-2*C_{6} - \mathcal{i}.(C_{4}+C_{8}) \f$
- * - channel #3 : \f$ 0.25 * (C_{1}+C_{3}+2*C_{6} \f$
- * - channel #4 : \f$ 0.25 * (C_{1}-C_{3}+2*C_{5}-2*C_{9} - 0.5\mathcal{i}.(C_{4}-C_{7}+C_{8}) \f$
- * - channel #5 : \f$ 0.25 * (C_{1}+C_{3}+4*C_{2}-2*C_{6}+4*C_{5}+4*C_{9}) \f$
+ * - channel #0 : \f$ 0.25 * (C33-i.\sqrt{2}.C23-C13+i.\sqrt{2}.C23^{*}-C13^{*}+2.C22-i.\sqrt{2}.C12+i.\sqrt{2}.C12^{*}+C11) \f$
+ * - channel #1 : \f$ 0.25 * (i.\sqrt{2}.C33+2.C23-i.\sqrt{2}.C13+i.\sqrt{2}.C13^{*}+2.C12^{*}-i.\sqrt{2}.C11) \f$
+ * - channel #2 : \f$ 0.25 * (-C33+i.\sqrt{2}.C23+C13+i.\sqrt{2}.C23^{*}+C13^{*}+2.C22-i.\sqrt{2}.C12-i.\sqrt{2}.C12^{*}-C11 ) \f$
+ * - channel #3 : \f$ 0.25 * (2.C33+2.C13+2.C13^{*}+2.C11)\f$
+ * - channel #4 : \f$ 0.25 * (i.\sqrt{2}.C33+i.\sqrt{2}.C13+2.C23^{*}-i.\sqrt{2}.C13^{*}+2.C12-i.\sqrt{2}.C11) \f$
+ * - channel #5 : \f$ 0.25 * (C33+i.\sqrt{2}.C23-C13-i.\sqrt{2}.C23^{*}-C13^{*}+2.C22+i.\sqrt{2}.C12-i.\sqrt{2}.C12^{*}+C11) \f$
*
- * Where:
- * - \f$ C_{1} = S_{hh}*S_{hh}}^{*} = \mathcal{Re}(input[0]) \f$
- * - \f$ C_{2} = S_{hv}*S_{hv}}^{*} = \mathcal{Re}(input[3]) \f$
- * - \f$ C_{3} = S_{vv}*S_{vv}}^{*} = \mathcal{Re}(input[5]) \f$
- * - \f$ C_{4} = \mathcal{Re}(S_{hh}*S_{hv}}^{*}) = \mathcal{Re}(input[1]) \f$
- * - \f$ C_{5} = \mathcal{Im}(S_{hh}*S_{hv}}^{*}) = \mathcal{Im}(input[1]) \f$
- * - \f$ C_{6} = \mathcal{Re}(S_{hh}*S_{vv}}^{*}) = \mathcal{Re}(input[2]) \f$
- * - \f$ C_{7} = \mathcal{Im}(S_{hh}*S_{vv}}^{*}) = \mathcal{Im}(input[2]) \f$
- * - \f$ C_{8} = \mathcal{Re}(S_{hv}*S_{vv}}^{*}) = \mathcal{Re}(input[4]) \f$
- * - \f$ C_{9} = \mathcal{Im}(S_{hv}*S_{vv}}^{*} = \mathcal{Im}(input[4])) \f$
+ * Where Cij are related to the elements of the reciprocal linear covariance matrix.
*
* The output pixel has 6 channels : the diagonal and the upper element of the reciprocal matrix.
* Element are stored from left to right, line by line.
@@ -75,51 +66,25 @@ public:
TOutput result;
result.SetSize(m_NumberOfComponentsPerPixel);
result.Fill(0.0);
-
- /*const RealType C1 = static_cast<RealType>(Covariance[0].real()); // C1 <hh.hh*>
- const RealType C2 = static_cast<RealType>(Covariance[3].real()); // C2 <hv.hv*>
- const RealType C3 = static_cast<RealType>(Covariance[5].real()); // C3 <vv.vv*>
- const RealType C4 = static_cast<RealType>(Covariance[1].real()); // C4 Re<hh.hv*>
- const RealType C5 = static_cast<RealType>(Covariance[1].imag()); // C5 Im<hh.hv*>
- const RealType C6 = static_cast<RealType>(Covariance[2].real()); // C6 Re<hh.vv*>
- const RealType C7 = static_cast<RealType>(Covariance[2].imag()); // C7 Im<hh.vv*>
- const RealType C8 = static_cast<RealType>(Covariance[4].real()); // C8 Re<hv.vv*>
- const RealType C9 = static_cast<RealType>(Covariance[4].imag()); // C9 Im<hv.vv*>
-
- const RealType llrrReal = 0.25 * ( C1 + C3 - 4*C2 -2*C6);
- const RealType llrrImag = -(C4 + C8);
-
- const RealType lllrReal = 0.25 * ( C1 - C3 - 2*C5 + 2*C9);
- const RealType lllrImag = -0.5*(C7+C4+C8);
-
- const RealType rrlrReal = 0.25 * ( C1 -C3 + 2*C5 - 2*C9);
- const RealType rrlrImag = 0.5 * (C4+C8-C7);
-
- result[0] = ComplexType(0.25 * ( C1 + C3 + 4*C2 - 2*C6 - 4*C5 - 4*C9));
- result[1] = ComplexType(lllrReal, lllrImag);
- result[2] = ComplexType(llrrReal, llrrImag);
- result[3] = ComplexType(0.25 * ( C1 + C3 + 2*C6));
- result[4] = ComplexType(rrlrReal, -rrlrImag);
- result[5] = ComplexType(0.25 * ( C1 + C3 + 4*C2 - 2*C6 + 4*C5 + 4*C9));*/
- const ComplexType A = static_cast<ComplexType>( Covariance[0] ); // <hh.hh*>
- const ComplexType B = static_cast<ComplexType>( Covariance[1] ); // <sqrt(2).hh.hv*>
- const ComplexType C = static_cast<ComplexType>( Covariance[2] ); // <hh.vv*>
- const ComplexType E = static_cast<ComplexType>( Covariance[3] ); // <2.hv.hv*>
- const ComplexType F = static_cast<ComplexType>( Covariance[4] ); // <sqrt(2).hv.vv*>
- const ComplexType I = static_cast<ComplexType>( Covariance[5] ); // <vv.vv*>
+ const ComplexType C11 = static_cast<ComplexType>( Covariance[0] ); // <hh.hh*>
+ const ComplexType C12 = static_cast<ComplexType>( Covariance[1] ); // <sqrt(2).hh.hv*>
+ const ComplexType C13 = static_cast<ComplexType>( Covariance[2] ); // <hh.vv*>
+ const ComplexType C22 = static_cast<ComplexType>( Covariance[3] ); // <2.hv.hv*>
+ const ComplexType C23 = static_cast<ComplexType>( Covariance[4] ); // <sqrt(2).hv.vv*>
+ const ComplexType C33 = static_cast<ComplexType>( Covariance[5] ); // <vv.vv*>
const ComplexType cst1 = ComplexType(0.0, vcl_sqrt(2.0));
const ComplexType two = ComplexType(2.0, 0 );
- result[0] = static_cast<ComplexType>( I-cst1*F-C+cst1*vcl_conj(F)-vcl_conj(C)+two*E-cst1*B+cst1*vcl_conj(B)+A ) ;
- result[1] = static_cast<ComplexType>( cst1*I+two*F-cst1*C+cst1*vcl_conj(C)+two*vcl_conj(B)-cst1*A );
- result[2] = static_cast<ComplexType>( -I+cst1*F+C+cst1*vcl_conj(F)+vcl_conj(C)+two*E-cst1*B-cst1*vcl_conj(B)-A ) ;
- result[3] = static_cast<ComplexType>( two*I+two*C+two*vcl_conj(C)+two*A ) ;
- result[4] = static_cast<ComplexType>( cst1*I+cst1*C+two*vcl_conj(F)-cst1*vcl_conj(C)+two*B-cst1*A ) ;
- result[5] = static_cast<ComplexType>( I+cst1*F-C-cst1*vcl_conj(F)-vcl_conj(C)+two*E+cst1*B-cst1*vcl_conj(B)+A ) ;
+ result[0] = static_cast<ComplexType>( C33-cst1*C23-C13+cst1*vcl_conj(C23)-vcl_conj(C13)+two*C22-cst1*C12+cst1*vcl_conj(C12)+C11 ) ;
+ result[1] = static_cast<ComplexType>( cst1*C33+two*C23-cst1*C13+cst1*vcl_conj(C13)+two*vcl_conj(C12)-cst1*C11 );
+ result[2] = static_cast<ComplexType>( -C33+cst1*C23+C13+cst1*vcl_conj(C23)+vcl_conj(C13)+two*C22-cst1*C12-cst1*vcl_conj(C12)-C11 ) ;
+ result[3] = static_cast<ComplexType>( two*C33+two*C13+two*vcl_conj(C13)+two*C11 ) ;
+ result[4] = static_cast<ComplexType>( cst1*C33+cst1*C13+two*vcl_conj(C23)-cst1*vcl_conj(C13)+two*C12-cst1*C11 ) ;
+ result[5] = static_cast<ComplexType>( C33+cst1*C23-C13-cst1*vcl_conj(C23)-vcl_conj(C13)+two*C22+cst1*C12-cst1*vcl_conj(C12)+C11 ) ;
result /= 4.0;
diff --git a/Modules/Filtering/Polarimetry/include/otbSinclairToReciprocalCircularCovarianceMatrixFunctor.h b/Modules/Filtering/Polarimetry/include/otbSinclairToReciprocalCircularCovarianceMatrixFunctor.h
index 6ae8b65f80..7dbb8322dc 100644
--- a/Modules/Filtering/Polarimetry/include/otbSinclairToReciprocalCircularCovarianceMatrixFunctor.h
+++ b/Modules/Filtering/Polarimetry/include/otbSinclairToReciprocalCircularCovarianceMatrixFunctor.h
@@ -83,17 +83,11 @@ public:
const ComplexType jS_hh = S_hh * ComplexType(0.0, 1.0);
const ComplexType jS_vv = S_vv * ComplexType(0.0, 1.0);
- /*const ComplexType Sll = coef * ( -S_hh-j2S_hv+S_vv );
- const ComplexType Slr = coef * ( -S_hh+-S_vv );
- const ComplexType Srr = coef * ( -S_hh+j2S_hv+S_vv );*/
const ComplexType Sll = coef * ( S_hh+j2S_hv-S_vv );
const ComplexType Slr = coef * ( jS_hh + jS_vv );
const ComplexType Srr = coef * ( -S_hh+j2S_hv+S_vv );
- //const ComplexType conjSll = vcl_conj(Sll);
- //const ComplexType conjSlr = vcl_conj(Slr);
- //const ComplexType conjSrr = vcl_conj(Srr);
SinclairToReciprocalCovarianceFunctorType funct;
return ( funct(Sll, Slr, Srr ) );
diff --git a/Modules/Filtering/Polarimetry/include/otbSinclairToReciprocalCovarianceMatrixFunctor.h b/Modules/Filtering/Polarimetry/include/otbSinclairToReciprocalCovarianceMatrixFunctor.h
index 07a616c792..8ad714abef 100644
--- a/Modules/Filtering/Polarimetry/include/otbSinclairToReciprocalCovarianceMatrixFunctor.h
+++ b/Modules/Filtering/Polarimetry/include/otbSinclairToReciprocalCovarianceMatrixFunctor.h
@@ -30,10 +30,10 @@ namespace Functor
*
* Output value are:
* - channel #0 : \f$ S_{hh}.S_{hh}^{*} \f$
- * - channel #1 : \f$ S_{hh}.S_{hv}^{*} \f$
+ * - channel #1 : \f$ \sqrt{2}.S_{hh}.S_{hv}^{*} \f$
* - channel #2 : \f$ S_{hh}.S_{vv}^{*} \f$
- * - channel #3 : \f$ S_{hv}.S_{hv}^{*} \f$
- * - channel #4 : \f$ S_{hv}.S_{vv}^{*} \f$
+ * - channel #3 : \f$ 2.S_{hv}.S_{hv}^{*} \f$
+ * - channel #4 : \f$ \sqrt{2}.S_{hv}.S_{vv}^{*} \f$
* - channel #5 : \f$ S_{vv}.S_{vv}^{*} \f$
*
* This is a adaptation of the SinclairToCovarianceMatrixFunctor, where \f$ S_{hv}=S_{vh} \f$.
@@ -71,13 +71,6 @@ public:
const ComplexType S_hh = static_cast<ComplexType>(Shh);
const ComplexType S_hv = static_cast<ComplexType>(Shv);
const ComplexType S_vv = static_cast<ComplexType>(Svv);
-
- /*result[0] = static_cast<OutputValueType>( std::norm( S_hh ) );
- result[1] = static_cast<OutputValueType>( S_hh*vcl_conj(S_hv) );
- result[2] = static_cast<OutputValueType>( S_hh*vcl_conj(S_vv) );
- result[3] = static_cast<OutputValueType>( std::norm( S_hv ) );
- result[4] = static_cast<OutputValueType>( S_hv*vcl_conj(S_vv) );
- result[5] = static_cast<OutputValueType>( std::norm( S_vv ) );*/
result[0] = static_cast<OutputValueType>( std::norm( S_hh ) );
result[1] = static_cast<OutputValueType>( vcl_sqrt(2.0)*S_hh*vcl_conj(S_hv) );
--
GitLab