3jISSKJr SSKrSSKJr SSKJrJr SSKJr S\RS\\ S\R4S jr SS \RS \RS \ S \ S\ S\ S\R4Sjjr"SS\R5rg))ListN)nn)filter2d_separableget_gaussian_kernel1d)_compute_paddingimgcropping_shapereturnc|[RRRXS*US*US*US*45$)z.Crop out the part of "valid" convolution area.r)torchr functionalpad)rr s M/home/wildlama/miniconda3/lib/python3.13/site-packages/kornia/metrics/ssim.py_croprsL 88   " " a  >!#4"4~a7H6H>Z[K\J\ ] img1img2 window_sizemax_valepspaddingc[U[R5(d[S[ U535e[U[R5(d[S[ U535e[U[ 5(d[S[ U535e[ UR5S:Xd[SUR35e[ UR5S:Xd[SUR35eURUR:Xd%[SURSUR35e[US URURS 9nS U-S -nS U-S -n[XU5n [XU5n /n US:Xa3URS=p[X/5n [X5n [X5n OUS:XaU S -nU S -nX-n[US -Xf5n[US -Xf5n[X-Xf5nUS:Xa%[UU 5n[UU 5n[UU 5nOUS:XaUU- nUU- nUU- nSU-U-SU-U--nX-U-UU-U--nUUU-- $)aCompute the Structural Similarity (SSIM) index map between two images. Measures the (SSIM) index between each element in the input `x` and target `y`. The index can be described as: .. math:: \text{SSIM}(x, y) = \frac{(2\mu_x\mu_y+c_1)(2\sigma_{xy}+c_2)} {(\mu_x^2+\mu_y^2+c_1)(\sigma_x^2+\sigma_y^2+c_2)} where: - :math:`c_1=(k_1 L)^2` and :math:`c_2=(k_2 L)^2` are two variables to stabilize the division with weak denominator. - :math:`L` is the dynamic range of the pixel-values (typically this is :math:`2^{\#\text{bits per pixel}}-1`). Args: img1: the first input image with shape :math:`(B, C, H, W)`. img2: the second input image with shape :math:`(B, C, H, W)`. window_size: the size of the gaussian kernel to smooth the images. max_val: the dynamic range of the images. eps: Small value for numerically stability when dividing. padding: ``'same'`` | ``'valid'``. Whether to only use the "valid" convolution area to compute SSIM to match the MATLAB implementation of original SSIM paper. Returns: The ssim index map with shape :math:`(B, C, H, W)`. Examples: >>> input1 = torch.rand(1, 4, 5, 5) >>> input2 = torch.rand(1, 4, 5, 5) >>> ssim_map = ssim(input1, input2, 5) # 1x4x5x5 z+Input img1 type is not a torch.Tensor. Got z+Input img2 type is not a torch.Tensor. Got z'Input max_val type is not a float. Got z,Invalid img1 shape, we expect BxCxHxW. Got: z,Invalid img2 shape, we expect BxCxHxW. Got: z,img1 and img2 shapes must be the same. Got: z and g?)devicedtypeg{Gz?r gQ?validsameg@) isinstancerTensor TypeErrortypefloatlenshape ValueErrorrrrrrr)rrrrrrkernelC1C2mu1mu2r heightwidthmu1_sqmu2_sqmu1_mu2 mu_img1_sq mu_img2_sq mu_img1_img2 sigma1_sq sigma2_sqsigma12numdens rssimr<"sV dELL ) )Ed4j\RSS dELL ) )Ed4j\RSS gu % %A$w-QRR tzz?a G |TUU tzz?a G |TUU :: #G |SXY]YcYcXdeff1c$++]a]g]ghFA%BA%B+4@C*4@C "N'b)))6/:C(C( F  !VF !VFiG#D!GV: F  V#IV#IW$Gw+g 0BCC2-)i2G"2LMC #) rc ^\rSrSrSrSS\S\S\S\SS4 U4S jjjrS \ RS \ RS\ R4S jr S r U=r $)SSIMaCreate a module that computes the Structural Similarity (SSIM) index between two images. Measures the (SSIM) index between each element in the input `x` and target `y`. The index can be described as: .. math:: \text{SSIM}(x, y) = \frac{(2\mu_x\mu_y+c_1)(2\sigma_{xy}+c_2)} {(\mu_x^2+\mu_y^2+c_1)(\sigma_x^2+\sigma_y^2+c_2)} where: - :math:`c_1=(k_1 L)^2` and :math:`c_2=(k_2 L)^2` are two variables to stabilize the division with weak denominator. - :math:`L` is the dynamic range of the pixel-values (typically this is :math:`2^{\#\text{bits per pixel}}-1`). Args: window_size: the size of the gaussian kernel to smooth the images. max_val: the dynamic range of the images. eps: Small value for numerically stability when dividing. padding: ``'same'`` | ``'valid'``. Whether to only use the "valid" convolution area to compute SSIM to match the MATLAB implementation of original SSIM paper. Shape: - Input: :math:`(B, C, H, W)`. - Target :math:`(B, C, H, W)`. - Output: :math:`(B, C, H, W)`. Examples: >>> input1 = torch.rand(1, 4, 5, 5) >>> input2 = torch.rand(1, 4, 5, 5) >>> ssim = SSIM(5) >>> ssim_map = ssim(input1, input2) # 1x4x5x5 rrrrr NcR>[TU]5 XlX lX0lX@lgN)super__init__rrrr)selfrrrr __class__s rrC SSIM.__init__s#  +%  rrrcp[XURURURUR5$rA)r<rrrr)rDrrs rforward SSIM.forwards'D 0 0$,,$,,WWr)rrrrg?g-q=r!)__name__ __module__ __qualname____firstlineno____doc__intr&strrCrr#rH__static_attributes__ __classcell__)rEs@rr>r>sh#JC%E\_mqXELLX XXXrr>rJ)typingrrrkornia.filtersrrkornia.filters.filterrr#rPrr&rQr<Moduler>rrrYs$ D2u||T#Y5<< i ,,i ,,ii i  i  i \\iX.X299.Xr