^j5dZddlmZddlZddlmZddlmZddlm Z ddl m Z ddl m Z ddlZddlZdd lmZmZmZmZmZdd lmZdd lmZdd lmZdd lmZgdZeddZ eeddZ!e ddZ"ddZ#d dZ$eed!dZ%d"dZ&d#d$dZ'd%d&dZ( d'dZ)d(dZ*d%d)dZ+d%d)dZ,d*dZ-y)+aOFunctional implementations of various blur operations for image processing. This module provides a collection of low-level functions for applying different blur effects to images, including standard blur, median blur, glass blur, defocus, and zoom effects. These functions form the foundation for the corresponding transform classes. ) annotationsN)Sequence)product)ceil)Literal)warn)clipped float32_iomaybe_process_in_chunkspreserve_channel_dimuint8_io)ValidationInfo)scale)convolve)EIGHT)box_blur central_zoomdefocus glass_blur median_blur zoom_blurcLttj||f}||S)zBlur an image. This function applies a blur to an image. Args: img (np.ndarray): Input image. ksize (int): Kernel size. Returns: np.ndarray: Blurred image. ksize)r cv2blurimgrblur_fns w/var/www/ramen.bs-engineer-server.com/venv/lib/python3.12/site-packages/albumentations/augmentations/blur/functional.pyrrs"&chhuenEG 3<cHttj|}||S)zMedian blur an image. This function applies a median blur to an image. Args: img (np.ndarray): Input image. ksize (int): Kernel size. Returns: np.ndarray: Median blurred image. r)r r medianBlurrs r rr/s&cnnEBG 3<r!c tjtj||d}|dk(rtj|j d|z |d}tj|j d|z |d}tj ||j d} tj||j d} t|D];} |dd| df} |dd| df} || | z| | zf|| | fc|| | f<|| | z| | zf<=n|dk(rttt|t|j d|z |dt|j d|z |dD]\\}\} } } |t|kr|n |t|z}||| df} ||| df} || | z| | zf|| | fc|| | f<|| | z| | zf<^ntd |d tj||dS) a|Glass blur an image. This function applies a glass blur to an image. Args: img (np.ndarray): Input image. sigma (float): Sigma. max_delta (int): Maximum delta. iterations (int): Number of iterations. dxy (np.ndarray): Dxy. mode (Literal["fast", "exact"]): Mode. Returns: np.ndarray: Glass blurred image. )rr)sigmaXrfastrNexactzUnsupported mode `z$`. Supports only `fast` and `exact`.) r GaussianBlurnparrayarangeshapetilerepeatrange enumeraterlen ValueError)rsigma max_delta iterationsdxymodexhswshwidydxindidxs r rrBs2 #uFCA v~ YYsyy|i/B ? YYsyy|i/B ? ggb"((1+6 iiBHHQK8z" DAQ1WBQ1WB)*1r61r6>):AadG &AadGQq2vq2v~& D ' j!ciilY. 2>ciilY. 2>  DNC!Qs3x#S3s8^CS!QYBS!QYB)*1r61r6>):AadG &AadGQq2vq2v~& D-dV3WXYY   Ae6 ::r!ctjtd| td|dz}|tkrdnd}tj||\}}tj |dz|dzz|dzktj }|tj|z}tj|||f|}t||S) zDefocus an image. This function defocuses an image. Args: img (np.ndarray): Input image. radius (int): Radius. alias_blur (float): Alias blur. Returns: np.ndarray: Defocused image. r(dtype)r%)kernel) r+r-maxrmeshgridr,float32sumrr*r) rradius alias_blurlengthrr:y aliased_diskrKs r rrzsYYAvAv(: ;F5AaE ;;vv &DAq88QTAqD[VQY6bjjILBFF<((L   lUEN: NF C ''r!cH|jdd\}}t||z t||z }}||z dz||z dz}}t||||z|||zf|tj}|jd|z dz|jd|z dz} }||||z| | |zfS)zCentral zoom an image. This function zooms an image. Args: img (np.ndarray): Input image. zoom_factor (int): Zoom factor. Returns: np.ndarray: Zoomed image. NrHrr()r.rrr INTER_LINEAR) r zoom_factorheightwidthh_chw_chh_topw_top h_trim_top w_trim_tops r rrsIIbqMMFEf{*+T%+2E-F$DTMa'%$,1)<5E EEDL(%%$,*>>?cN^N^ _C!iilV39CIIaL5P1PP QQr!ctj|tj}|D]}|t||z }||zt |dzz S)zZoom blur an image. This function zooms and blurs an image. Args: img (np.ndarray): Input image. zoom_factors (np.ndarray | Sequence[int]): Zoom factors. Returns: np.ndarray: Zoomed and blurred image. rIr()r+ zeros_likerNrr3)r zoom_factorsoutrWs r rrsS --2:: .C#.  |C--. #I#l+a/ 00r!c |d|ks|d|kr>t||dt||df}t|d|d|d|dtd|S|S) Nrr(z: Invalid kernel size range z. Values less than z2 are not allowed. Range automatically adjusted to .rH stacklevel)rLr UserWarning)result min_value field_name new_results r _ensure_min_valuerms} ay9q I 5)VAY/Yq 1JK  l6vh? ){+//9l! =    Mr!c|ddk(s |ddzdk(r|dn|ddz|ddk(s |ddzdk(r|dn|ddzf}||k7rt|d|d|dtd|S)NrrHr(z+: Non-zero kernel sizes must be odd. Range z automatically adjusted to rerf)rrh)rirkrls r _ensure_odd_valuesrosAY!^vay1}'9q vay1}AY!^vay1}'9q vay1}JV lEfXMhishttu v  r!c~t|tr6t|dk7r tdt |dt |df}n |t |f}t |||j }t||j }|d|dkDr1|d|df}t|j d|d|dtd|S|S) z0Process blur limit to ensure valid kernel sizes.rHz(Sequence must contain exactly 2 elementsrr(z: Invalid range z. (min > max). Range automatically adjusted to rerf) isinstancerr3r4intrmrkrorrh)valueinforjri final_results r process_blur_limitrvs%" u:?GH HeAh-U1X/SZ( vy$// BF  8F ay6!9q 6!9-  /x7efresst u   Mr!c:tj|dd}tj||ftj}|dz}tj|}tj |}tj |} |dz} |dkr t|} t| } | d| z z} n3|dkDr|} | d| z z} t| } nt| } t| } tj| | |}|||zz}|| |zz}|r:|jdd| zdz }|jdd| zdz }||z }||z }tjtj|d|dz jt}tjtj|d|dz jt}tjtj||gd}d||d d df|d d dff<|j!sd|||f<|S) aCreate a motion blur kernel. Args: kernel_size (int): Size of the kernel (must be odd) angle (float): Angle in degrees (counter-clockwise) direction (float): Blur direction (-1.0 to 1.0) allow_shifted (bool): Allow kernel to be randomly shifted from center random_state (random.Random): Python's random.Random instance Returns: np.ndarray: Motion blur kernel gg?rIrHrr(r')axisN)r+clipzerosrNdeg2radcossinabsfloatlinspaceuniformroundastyperrunique column_stackany) kernel_sizeangle direction allow_shifted random_staterKcenter angle_radrAr@ line_length bias_factort_startt_endtr:rSshift_xshift_ypointss r create_motion_kernelrs* 4-I XX{K0 CF A F 5!I  B  B"K1})n  %q;/ Q ,!k/2k" %k" GUK0A aAaA&&r1- ;a?&&r1- ;a? W  W   Q a077= low rFrHrr()rLrandint)rlowhighnum_odd_valuesrand_idxs r sample_odd_from_rangerIs& aa1 % &C q$$(a-( )D tS>D d{ SjQ&*N##A~'9:H !h, r!cR|dk(rt|dzdzdzn|}|dzdk(r|dzn|}tj|dz |dz|}tjd||z dzz}||j z }|ddtj f|tj ddfzS)akCreate a Gaussian kernel following PIL's approach. Args: sigma (float): Standard deviation for Gaussian kernel. ksize (int): Kernel size. If 0, size is computed as int(sigma * 3.5) * 2 + 1 to match PIL's implementation. Otherwise, must be positive and odd. Returns: np.ndarray: 2D normalized Gaussian kernel. r @rHr(N)rrr+rexprOnewaxisr5rsizer: kernel_1ds r create_gaussian_kernelrms(-z3us{ a ! #uDax1}4!8$D daiL$!)T2Atq5yQ../IIMMO+I Q ] #i A &> >>r!c|dk(rt|dzdzdzn|}|dzdk(r|dzn|}t|}tjd||z dzz}||j z S)anCreate a 1D Gaussian kernel following PIL's approach. Args: sigma (float): Standard deviation for Gaussian kernel. ksize (int): Kernel size. If 0, size is computed as int(sigma * 3.5) * 2 + 1 to match PIL's implementation. Otherwise, must be positive and odd. Returns: np.ndarray: 1D normalized Gaussian kernel. rrrHr(rr)rr"create_gaussian_kernel_input_arrayr+rrOrs r create_gaussian_kernel_1drsv(-z3us{ a ! #uDax1}4!8$D +5Atq5yQ../I y}} &&r!c |dkr3tjtt|dz |dzdzdStj|dz |dz|S)aCreates a 1-D array which will create an array of x-coordinates which will be input for the gaussian function (values from -size/2 to size/2 with step size of 1) Piecewise function is needed as equivalent python list comprehension is faster than np.linspace for values of size < 100 Args: size (int): kernel size Returns: np.ndarray: x-coordinate array which will be input for gaussian function that will be used for separable gaussian blur drHr()r+r,listr1rrs r rrsS czxxUTQY<$!)q!DEFF ;;|TQY 55r!)r np.ndarrayrrrreturnr)rrr5rr6rrr7rrr8rr9zLiteral['fast', 'exact']rr)rrrPrrrQrrr)rrrWrrrr)rrrbznp.ndarray | Sequence[int]rr)rituple[int, int]rjrrrk str | Nonerr)N)rirrkrrr)r)rszint | tuple[int, int]rtrrjrrrr) rrrrrrrrboolr random.Randomrr)rrrrrrrrrrr)r5rrrrrr)rrrrr).__doc__ __future__rrandomcollections.abcr itertoolsrmathrtypingrwarningsrrnumpyr+albucorer r r r r pydanticr1albumentations.augmentations.geometric.functionalr-albumentations.augmentations.pixel.functionalr$albumentations.core.type_definitionsr__all__rrrrrrrmrorvrrrrrr!r rsU# $ aa#CB6 ["  "4; 4; 4;4; 4;  4; # 4;4;4;n(4R, 1  1*  2PP PP P  P  Pf! H?:'46r!