o oi*@sddlmZddlZddlmZddlmZmZddlm Z m Z gdZ d(d d Z d)d dZ d*ddZd+ddZd,d-ddZd.d/d!d"Zd.d0d#d$Zd1d2d&d'ZeZdS)3) annotationsN)Tensor)KORNIA_CHECK_IS_TENSORKORNIA_CHECK_SHAPE)convert_points_from_homogeneousconvert_points_to_homogeneous) batched_dot_productbatched_squared_normcompose_transformationseuclidean_distanceinverse_transformationpoint_line_distancerelative_transformation squared_normtransform_pointstrans_01rtrans_12returnc CsPt|t||dvr|jdddkstd|j|dvr.|jdddks6td|j||krLtd|d||d dd dd f}|d dd dd f}|d dd d df}|d dd d df}t||}t|||}||j}||d dd dd f<||d dd d df<d |d <|S) alCompose two homogeneous transformations. .. math:: T_0^{2} = \begin{bmatrix} R_0^1 R_1^{2} & R_0^{1} t_1^{2} + t_0^{1} \ \\mathbf{0} & 1\end{bmatrix} Args: trans_01: tensor with the homogeneous transformation from a reference frame 1 respect to a frame 0. The tensor has must have a shape of :math:`(N, 4, 4)` or :math:`(4, 4)`. trans_12: tensor with the homogeneous transformation from a reference frame 2 respect to a frame 1. The tensor has must have a shape of :math:`(N, 4, 4)` or :math:`(4, 4)`. Returns: the transformation between the two frames with shape :math:`(N, 4, 4)` or :math:`(4, 4)`. Example:: >>> trans_01 = torch.eye(4) # 4x4 >>> trans_12 = torch.eye(4) # 4x4 >>> trans_02 = compose_transformations(trans_01, trans_12) # 4x4 Nrz8Input trans_01 must be a of the shape Nx4x4 or 4x4. Got z8Input trans_12 must be a of the shape Nx4x4 or 4x4. Got %Input number of dims must match. Got  and .r?.rr)rdimshape ValueErrortorchmatmul new_zeros) rrrmat_01rmat_12tvec_01tvec_12rmat_02tvec_02trans_02r+J/home/ubuntu/.local/lib/python3.10/site-packages/kornia/geometry/linalg.pyr 's&  r cCst||dvr|jdddkstd|j|dddddf}|dddddf}|d d}t| |}||j}|dddddf||dddddf|d |d <|S) aInvert a 4x4 homogeneous transformation. :math:`T_1^{2} = \begin{bmatrix} R_1 & t_1 \\ \mathbf{0} & 1 \end{bmatrix}` The inverse transformation is computed as follows: .. math:: T_2^{1} = (T_1^{2})^{-1} = \begin{bmatrix} R_1^T & -R_1^T t_1 \\ \mathbf{0} & 1\end{bmatrix} Args: trans_12: transformation tensor of shape :math:`(N, 4, 4)` or :math:`(4, 4)`. Returns: tensor with inverted transformations with shape :math:`(N, 4, 4)` or :math:`(4, 4)`. Example: >>> trans_12 = torch.rand(1, 4, 4) # Nx4x4 >>> trans_21 = inverse_transformation(trans_12) # Nx4x4 rrNrz'Input size must be a Nx4x4 or 4x4. Got .rrrr) rrrr transposer!r"r#copy_)rr%r'rmat_21tvec_21trans_21r+r+r,r \s  r r*c CsZt|t||dvr|jdddkstd|j|dvr.|jdddks6td|j||ksLtd|d||ddd dd f}|ddd d d f}|ddd dd f}|ddd d d f}|d d}t||}t|||}t|} || ddd dd f<|| ddd d d f<d | d <| S)ayCompute the relative homogeneous transformation from a reference transformation. :math:`T_1^{0} = \begin{bmatrix} R_1 & t_1 \\ \mathbf{0} & 1 \end{bmatrix}` to destination :math:`T_2^{0} = \begin{bmatrix} R_2 & t_2 \\ \mathbf{0} & 1 \end{bmatrix}`. The relative transformation is computed as follows: .. math:: T_1^{2} = (T_0^{1})^{-1} \cdot T_0^{2} Args: trans_01: reference transformation tensor of shape :math:`(N, 4, 4)` or :math:`(4, 4)`. trans_02: destination transformation tensor of shape :math:`(N, 4, 4)` or :math:`(4, 4)`. Returns: the relative transformation between the transformations with shape :math:`(N, 4, 4)` or :math:`(4, 4)`. Example:: >>> trans_01 = torch.eye(4) # 4x4 >>> trans_02 = torch.eye(4) # 4x4 >>> trans_12 = relative_transformation(trans_01, trans_02) # 4x4 rrNrz/Input must be a of the shape Nx4x4 or 4x4. Got rr.rrr-rr)rrrr r.r!r" zeros_like) rr*r$r&r(r)rmat_10r%r'rr+r+r,rs(   rpoints_1cCs6t|t||jd|jdks%|jddkr%td|jd|j|jd|jddks;td|d|t|j}|d|jd|jd}|d|jd|jd}tj|t|jd|jddd}t|}t || dd d}tj |dd }t |}|jd|d<|jd|d<||}|S) aApply transformations to a set of points. Args: trans_01: tensor for transformations of shape :math:`(B, D+1, D+1)`. points_1: tensor of points of shape :math:`(B, N, D)`. Returns: a tensor of N-dimensional points. Shape: - Output: :math:`(B, N, D)` Examples: >>> points_1 = torch.rand(2, 4, 3) # BxNx3 >>> trans_01 = torch.eye(4).view(1, 4, 4) # Bx4x4 >>> points_0 = transform_points(trans_01, points_1) # BxNx3 rz=Input batch size must be the same for both tensors or 1. Got rr-z1Last input dimensions must differ by one unit Gotr)repeatsrr)r) rrr listreshaper!repeat_interleaveintrbmmpermutesqueezer)rr5 shape_inp points_1_h points_0_hpoints_0r+r+r,rs(" $ r& .>pointlineepsfloatcCst|t||jddvrtd|j|jddkr&td|j|d|d}||d|d7}||d7}||d|d}|||S) a^Return the distance from points to lines. Args: point: (possibly homogeneous) points :math:`(*, N, 2 or 3)`. line: lines coefficients :math:`(a, b, c)` with shape :math:`(*, N, 3)`, where :math:`ax + by + c = 0`. eps: Small constant for safe sqrt. Returns: the computed distance with shape :math:`(*, N)`. r-rz&pts must be a (*, 2 or 3) tensor. Got rz#lines must be a (*, 3) tensor. Got ).r).r6).r)rrr abs_squaresqrt)rDrErF numerator denom_normr+r+r,r s   r FxykeepdimboolcCs,t|ddgt|ddg||d|S)z#Return a batched version of .dot().*Nr-)rsum)rMrNrOr+r+r,r srcCs t|||S)z$Return the squared norm of a vector.)r)rMrOr+r+r,r s r ư>cCs>t|ddgt|ddg||djd|d|S)a{Compute the Euclidean distance between two set of n-dimensional points. More: https://en.wikipedia.org/wiki/Euclidean_distance Args: x: first set of points of shape :math:`(*, N)`. y: second set of points of shape :math:`(*, N)`. keepdim: whether to keep the dimension after reduction. eps: small value to have numerical stability. rQrRrr-)rrO)rpowrSadd_sqrt_)rMrNrOrFr+r+r,r s "r )rrrrrr)rrrr)rrr*rrr)rrr5rrr)rC)rDrrErrFrGrr)F)rMrrNrrOrPrr)rMrrOrPrr)FrT) rMrrNrrOrPrFrGrr) __future__rr! kornia.corerkornia.core.checkrrkornia.geometry.conversionsrr__all__r r rrr rr r rr+r+r+r,s   5 + 1 2 !