3jk@ <SSKrSSKJrJr SSKrSSKJr SSKJrJ r J r J r SSK J r Jr SSKJr SSKJr \\R&\R&4rSS\R&S \R&S \R&S \S \S \R&4 SjjrSS\R&S \R&S \R&S \S \S \R&4 SjjrSS\R&S\R&S \R&S \S \R&4 SjjrS S\R&S\R&S\\R&S\S \R&4 SjjrS!S\R&S\R&S\R&S\S\S \R&4 SjjrS\R&S\R&S \R&4SjrS"S\R&S\R&S\\R&S \R&4SjjrS#S\R&S\R&S\R&S\S\S \R&4 Sjjr g)$N)OptionalTuple)KORNIA_CHECK_SHAPE)_extract_device_dtype_torch_svd_castsafe_inverse_with_masksafe_solve_with_mask)convert_points_from_homogeneousconvert_points_to_homogeneous)normalize_points)transform_pointspts1pts2Hsquaredepsreturnc[U/SQ5 URSS:Xa[U5nUSnUSnOUSSS2S4nUSSS2S 4nURSS:Xa[U5nUSn USn OUSSS2S4n USSS2S 4n US S n US S n US S n USS nUSS nUSS nUSS nUSS nUSS nX-X--U -nX-X--U-nUU-UU--U-nUUU-- nUUU-- nUU - RS5UU - RS5-nU(aU$UU-R 5$)a9Return transfer error in image 2 for correspondences given the homography matrix. Args: pts1: correspondences from the left images with shape (B, N, 2 or 3). If they are homogeneous, converted automatically. pts2: correspondences from the right images with shape (B, N, 2 or 3). If they are homogeneous, converted automatically. H: Homographies with shape :math:`(B, 3, 3)`. squared: if True (default), the squared distance is returned. eps: Small constant for safe sqrt. Returns: the computed distance with shape :math:`(B, N)`. B3r.r..Nrr).rr).N).rr).r).rr).rr).rr).rr).rr).rrr)rshaper powsqrt)rrrrrx1y1x1y1u2v2u2v2h00h01h02h10h11h12h20h21h22x_numy_numw_denu1in2v1in2err2s T/home/wildlama/miniconda3/lib/python3.13/site-packages/kornia/geometry/homography.pyoneway_transfer_errorr7 s$q/* zz"~.t4 &\ &\ #q!)_ #q!)_ zz"~.t4 &\ &\ #q!)_ #q!)_ I,y !C I,y !C I,y !C I,y !C I,y !C I,y !C I,y !C I,y !C I,y !C Hsx # %E Hsx # %E "HsRx # %E US[ !E US[ !E BJ  A %"*!1!1!!4 4D 3J   c:[U/SQ5 URS5S:Xa [U5nURS5S:Xa [U5n[R"UR 5R n[U5upg[XUSU5n[XUSU5n URSS5RU5n X-U RUR 5-XZ)RUR 5--n U(aU $X-R5$)a8Return Symmetric transfer error for correspondences given the homography matrix. Args: pts1: correspondences from the left images with shape (B, N, 2 or 3). If they are homogeneous, converted automatically. pts2: correspondences from the right images with shape (B, N, 2 or 3). If they are homogeneous, converted automatically. H: Homographies with shape :math:`(B, 3, 3)`. squared: if True (default), the squared distance is returned. eps: Small constant for safe sqrt. Returns: the computed distance with shape :math:`(B, N)`. rrrTr) rsizer torchfinfodtypemaxrr7view expand_astor ) rrrrrmax_numH_invgood_Htherebackgood_H_reshapeouts r6symmetric_transfer_errorrIas$q/* yy}.t4 yy}.t4kk$**%))G+1-ME/AtSIE.t5$LD#);;r1#5#?#?#FN <>,,U[[9 9GFZFZ[`[f[fFg>@G 6G# $Eq Lr8points1points2weightssolverc PURUR:wa[UR5eURSS:a[UR5e[U/SQ5 [U/SQ5 [X/5upESn[ U5upx[ U5up[ R "USSS9up[ R "U SSS9up[ R"U 5[ R"U 5nn[ R"UUUU *U *U*X-X-U/ SS9n[ R"XUUUUU *U -U *U -U */ SS9n[ R"UU4SS9RURS SURS5nUcURS S5U-nO[UR5S:XaURURS S:Xd[UR5eURSSS9RS5nURS S5U-U-nUS :Xa'[U5u nnUSR-SSS5nO`US:XaT[ R."URS URSXES9n[1UU5un nURSSS5nO[2e[5U 5S UU--nUUSSS 2SS 24U-- nU$![ aD ["R$"S [&SS9 [ R("UR+S 5SS4XES9s$f=f)a%Compute the homography matrix using the DLT formulation. The linear system is solved by using the Weighted Least Squares Solution for the 4 Points algorithm. Args: points1: A set of points in the first image with a tensor shape :math:`(B, N, 2)`. points2: A set of points in the second image with a tensor shape :math:`(B, N, 2)`. weights: Tensor containing the weights per point correspondence with a shape of :math:`(B, N)`. solver: variants: svd, lu. Returns: the computed homography matrix with shape :math:`(B, 3, 3)`. r)rrNrO:0yE>rrrPrSrNrTsvdSVD did not converge stacklevelrdevicer=.rlu.)rAssertionErrorrrr r;rU ones_like zeros_likecatreshaperXlenrepeat_interleave unsqueezer RuntimeErrorwarningswarnRuntimeWarningemptyr:r?onesr NotImplementedErrorr)rgrhrirjrsr=r points1_norm transform1 points2_norm transform2r"r#x2y2rzerosaxayAw_full_VrrsolH_norms r6find_homography_dltrs$}} %W]]++}}Q!W]]++w0w0)7*<=MFC/8L/8L [[2a 8FB [[2a 8FB//"%u'7'7';%D E5%"rcD5"'27BOUW XB BD%sRx"rB3OUW XB 2r(#++BHHQKRXXb\JA KKB ! #GMM"a'GMMW]]2A=N,N / /**1!*4>>qA [[R 6 )Q .  Y%a(GAq! gJOOB1 % 4 JJqwwqz1771:f J(A. Q KKAq !!!z*1-Z@A !CbcM"S( )F M Y MM0.Q O;; 1 1! 4a;FX X Ys)KA L%$L% soft_inl_thn_iterc[XU5n[US- 5H<n[XUS5n[R"U*SUS--- 5n[XU5nM> U$)aCompute the homography matrix using the iteratively-reweighted least squares (IRWLS). The linear system is solved by using the Reweighted Least Squares Solution for the 4 Points algorithm. Args: points1: A set of points in the first image with a tensor shape :math:`(B, N, 2)`. points2: A set of points in the second image with a tensor shape :math:`(B, N, 2)`. weights: Tensor containing the weights per point correspondence with a shape of :math:`(B, N)`. Used for the first iteration of the IRWLS. soft_inl_th: Soft inlier threshold used for weight calculation. n_iter: number of iterations. Returns: the computed homography matrix with shape :math:`(B, 3, 3)`. rF@r)rrangerIr;exp) rgrhrirrrrerrors weights_news r6find_homography_dlt_iteratedrsb&*'GDA 6A: 7!US$)IIvg Q9O.P$Q + > Hr8cURUR:wa[UR5e[R"/SQURS9n[R"/SQURS9n[R"/SQURS9nUSS2U4USS2U4USS2U4pvnUSS2U4USS2U4USS2U4pnS[R S[R S[R S [R 4S jn [R "U "XVU55n [R "U "XU 55n X:HRS S 9nU$) aImplement oriented constraint check from :cite:`Marquez-Neila2015`. Analogous to https://github.com/opencv/opencv/blob/4.x/modules/calib3d/src/usac/degeneracy.cpp#L88 Args: points1: A set of points in the first image with a tensor shape :math:`(B, 4, 2)`. points2: A set of points in the second image with a tensor shape :math:`(B, 4, 2)`. Returns: Mask with the minimal sample is good for homography estimation :math:`(B)`. )rrrr)rs)rrrr)rrrrNabcrc@X- nX - nUSUS-USUS-- $)Nrr)rrrabacs r6_orient/sample_is_valid_for_homography.._orient(s5 U U&zBvJ&Fbj)@@@r8rrS)rrvr;tensorrsTensorsignall)rgrhidx_iJKp1_ip1_jp1_kp2_ip2_jp2_kr left_sign right_signsample_is_valids r6sample_is_valid_for_homographyrs)}} %W]]++ LLgnn =E \'..9A \'..9Aq%x('!Q$-ADq%x('!Q$-ADA5<<AELLAU\\AellA  74t45IGD56J!.333:O r8c [UR5S:XaUSn[UR5S:XaUSn[U/SQ5 [U/SQ5 URSSup4[X/5upVUR USU-S5nUR USU-S5n[ U5up[ U5up[ R"U SSS9up[ R"U SSS9unn[ R"U SSS9unn[ R"USSS9unn[ R"USSS9unn[ R"USSS9unnUU- nUU- nUU-UU-- nSn[ R"UU-UU-UUU-UU-UUU-UU-U/ SS 9n[ R"UU-UU-UUU-UU-UUU-UU-U/ SS 9n[ R"UU4SS 9R URS SURS5nUcURS S5U-nO[UR5S:XaURURSS:Xd[UR5e[ R"URSS 9RSSS5R URS S55nURS S5U-U-n[U5u n n!U!SR+SSS5n"[-U 5S U"U --n"U"U"SSS2SS24U-- n#U#$![aD [ R""S [$SS 9 [ R&"U R)S 5SS4XVS9s$f=f)a`Compute the homography matrix using the DLT formulation for line correspondences. See :cite:`homolines2001` for details. The linear system is solved by using the Weighted Least Squares Solution for the 4 Line correspondences algorithm. Args: ls1: A set of line segments in the first image with a tensor shape :math:`(B, N, 2, 2)`. ls2: A set of line segments in the second image with a tensor shape :math:`(B, N, 2, 2)`. weights: Tensor containing the weights per point correspondence with a shape of :math:`(B, N)`. Returns: the computed homography matrix with shape :math:`(B, 3, 3)`. rNrMrrrPrrmrSrrTrorprrrt.)r{rrrrzr r;rUryrXrv diag_embedr}repeatrr~rrrrr:r?r)$rJrKriBSrNrsr=rgrhrrrrlst1le1lst2le2xs1ys1xs2ys2xe1ye1xe2ye2rrCrrrw_diagrrrrs$ r6find_homography_lines_dltr5sn$ 399~$i 399~$is01s01 IIbqMEB)3*5MFkk"a!eQ'Gkk"a!eQ'G/8L/8L La:ID La:ID#{{4R2HC{{4R2HC{{3Bq1HC{{3Bq1HC c A c A c C#IAC AGQWaS!c'1a#gq3wPQRXZ [B AGQWaS!c'1a#gq3wPQRXZ [B 2r(#++BHHQKRXXb\JA KKB ! #W]]#q(w}} "1 /M / /!!'"3"3"3";"B"B1a"K"S"ST[TaTabcTdfh"ij KKB & (1 ,U!!$1a ' Aq!Az*1-Z@A !CbcM"S( )F M U ,nK{{L--a0!Q7TTUsL''A M54M5c[XU5n[US- 5H<n[XUS5n[R"U*SUS--- 5n[XU5nM> U$)aCompute the homography matrix using the iteratively-reweighted least squares (IRWLS) from line segments. The linear system is solved by using the Reweighted Least Squares Solution for the 4 line segments algorithm. Args: ls1: A set of line segments in the first image with a tensor shape :math:`(B, N, 2, 2)`. ls2: A set of line segments in the second image with a tensor shape :math:`(B, N, 2, 2)`. weights: Tensor containing the weights per point correspondence with a shape of :math:`(B, N)`. Used for the first iteration of the IRWLS. soft_inl_th: Soft inlier threshold used for weight calculation. n_iter: number of iterations. Returns: the computed homography matrix with shape :math:`(B, 3, 3)`. rFrr)rrrfr;r) rJrKrirrrrrrs r6"find_homography_lines_dlt_iteratedr~sc&0'BA 6A: B3QPUV$)IIvg Q9O.P$Q %c < Hr8)Trm)F)Nru)g@)N)g@r)!rtypingrrr;kornia.core.checkrkornia.core.utilsrrrr kornia.geometry.conversionsr r kornia.geometry.epipolarr kornia.geometry.linalgr r TupleTensorboolfloatr7rIrfstrrintrrrrrr8r6rsq$" 0rrf53ELL%,,./ ae> ,,>#ll>/4||>FJ>X]> \\>Dae$ ,,$#ll$/4||$FJ$X]$ \\$PLQ$ $!LL$-2\\$DH$ \\$PimA \\A$)LLA;CELL;QAbeA \\AJrs \\ $)LL ;@<< V[ kn  \\ 6$ELL$5<<$TYT`T`$PMQF F!LLF3;ELL3IF \\FTjk  !LL 38<< NS cf  \\ r8