o :ήc@sdZddlZddlmZddlmZmZm Z ddl m Z ddl m Z gd Z  dAd d Z  dAd d Z dBddZ dBddZ dCddZeddiddd  dDddZeddiddd  dDddZeddiddd  dDddZeddiddd  dDd d!Zeddiddd  dDd"d#Zeddiddd  dDd$d%Zeddiddd  dDd&d'Zeddiddd  dEd(d)Zeddiddd  dDd*d+Zeddiddd  dDd,d-Zeddiddd  dDd.d/Zeddiddd  dDd0d1Zeddiddd  dDd2d3Z eddiddd  dDd4d5Z!eddiddd  dDd6d7Z"eddiddd  dDd8d9Z#eddiddd  dDd:d;Z$eddiddd  dFdd?d@Z&dS)Gu General Description ------------------- These filters compute the local histogram at each pixel, using a sliding window similar to the method described in [1]_. A histogram is built using a moving window in order to limit redundant computation. The moving window follows a snake-like path: ...------------------------↘ ↙--------------------------↙ ↘--------------------------... The local histogram is updated at each pixel as the footprint window moves by, i.e. only those pixels entering and leaving the footprint update the local histogram. The histogram size is 8-bit (256 bins) for 8-bit images and 2- to 16-bit for 16-bit images depending on the maximum value of the image. The filter is applied up to the image border, the neighborhood used is adjusted accordingly. The user may provide a mask image (same size as input image) where non zero values are the part of the image participating in the histogram computation. By default the entire image is filtered. This implementation outperforms :func:`skimage.morphology.dilation` for large footprints. Input images will be cast in unsigned 8-bit integer or unsigned 16-bit integer if necessary. The number of histogram bins is then determined from the maximum value present in the image. Eventually, the output image is cast in the input dtype, or the `output_dtype` if set. To do ----- * add simple examples, adapt documentation on existing examples * add/check existing doc * adapting tests for each type of filter References ---------- .. [1] Huang, T. ,Yang, G. ; Tang, G.. "A fast two-dimensional median filtering algorithm", IEEE Transactions on Acoustics, Speech and Signal Processing, Feb 1979. Volume: 27 , Issue: 1, Page(s): 13 - 18. N)ndimage)check_nDdeprecate_kwargwarn) img_as_ubyte) generic_cy) autolevelequalizegradientmaximummeangeometric_mean subtract_meanmedianminimummodalenhance_contrastpop threshold noise_filterentropyotsuc Csnt|d|j}|ttfvs|ttfvrtd|tjtjfvr0d|d}t|ddt|}t t|dk}|j |j krCtdt |}|d urUt|}t |}||ur]t d |d urt|d urh|j}tj |j |f|d }nt|j dkr||j |f}|jtjtjfvrd }n ttd |d}|dkrtd|dt|dddd|||||fS)a&Preprocess and verify input for filters.rank methods. Parameters ---------- image : 2-D array (integer or float) Input image. footprint : 2-D array (integer or float), optional The neighborhood expressed as a 2-D array of 1's and 0's. out : 2-D array (integer or float), optional If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). out_dtype : data-type, optional Desired output data-type. Default is None, which means we cast output in input dtype. pixel_size : int, optional Dimension of each pixel. Default value is 1. Returns ------- image : 2-D array (np.uint8 or np.uint16) footprint : 2-D array (np.uint8) The neighborhood expressed as a binary 2-D array. out : 3-D array (same dtype out_dtype or as input) Output array. The two first dimensions are the spatial ones, the third one is the pixel vector (length 1 by default). mask : 2-D array (np.uint8) Mask array that defines (>0) area of the image included in the local neighborhood. n_bins : int Number of histogram bins. zdtype cannot be bool.1Possible precision loss converting image of type p to uint8 as required by rank filters. Convert manually using skimage.util.img_as_ubyte to silence this warning. stacklevelr8Image dimensions and neighborhood dimensionsdo not matchN'Cannot perform rank operation in place.dtyperrGBad rank filter performance is expected due to a large number of bins (,), equivalent to an approximate bitdepth of .1f.)rr#bool ValueErrornpuint8uint16rrascontiguousarrayndimNotImplementedErroremptyshapelenreshapeint8intmaxlog2) image footprintoutmask out_dtype pixel_size input_dtypemessagen_binsrCC/tmp/pip-target-vg8gfxp4/lib/python/skimage/filters/rank/generic.py_preprocess_inputAsJ $     rEc CsVt|d|jtjtjfvrd|jd}t|ddt|}tt|dk}|j|jkr2t dt|}|durEtj |j tjd }n t|}t|}||urVt d |durm|dura|j}tj |j |f|d }n ||j |f}|jtjtjfv}|rd }n ttd|d }|d krtd|dt|dddd|||||fS)a&Preprocess and verify input for filters.rank methods. Parameters ---------- image : 3-D array (integer or float) Input image. footprint : 3-D array (integer or float), optional The neighborhood expressed as a 3-D array of 1's and 0's. out : 3-D array (integer or float), optional If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). out_dtype : data-type, optional Desired output data-type. Default is None, which means we cast output in input dtype. pixel_size : int, optional Dimension of each pixel. Default value is 1. Returns ------- image : 3-D array (np.uint8 or np.uint16) footprint : 3-D array (np.uint8) The neighborhood expressed as a binary 3-D array. out : 3-D array (same dtype out_dtype or as input) Output array. The two first dimensions are the spatial ones, the third one is the pixel vector (length 1 by default). mask : 3-D array (np.uint8) Mask array that defines (>0) area of the image included in the local neighborhood. n_bins : int Number of histogram bins. rrrrrrr Nr"r!r$rr%r&r'r(r))rr#r,r-r.rrr/r0r+onesr3r1r2r5r6r7r8r9) r:r;r<r=r>r?rAis_8bitrBrCrCrD_handle_input_3DsF $     rHc Cs>t|||||\}}}}}||||||||dtj|ddS)azProcess the specific cython function to the image. Parameters ---------- func : function Cython function to apply. image : 2-D array (integer or float) Input image. footprint : 2-D array (integer or float) The neighborhood expressed as a 2-D array of 1's and 0's. out : 2-D array (integer or float) If None, a new array is allocated. mask : ndarray (integer or float) Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). out_dtype : data-type, optional Desired output data-type. Default is None, which means we cast output in input dtype. shift_xshift_yr=r<rB)axis)rEr,squeeze) funcr:r;r<r=rJrKr>rBrCrCrD_apply_scalar_per_pixels rPc CsFt|||||\}}}}} ||||||||| d||jddS)N)rJrKshift_zr=r<rBr)rHr5r3) rOr:r;r<r=rJrKrQr>rBrCrCrD_apply_scalar_per_pixel_3Ds  rRc Cs6t||||||\}}}}} |||||||| d|S)a| Parameters ---------- func : function Cython function to apply. image : 2-D array (integer or float) Input image. footprint : 2-D array (integer or float) The neighborhood expressed as a 2-D array of 1's and 0's. out : 2-D array (integer or float) If None, a new array is allocated. mask : ndarray (integer or float) Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). out_dtype : data-type, optional Desired output data-type. Default is None, which means we cast output in input dtype. pixel_size : int, optional Dimension of each pixel. Returns ------- out : 3-D array with float dtype of dimensions (H,W,N), where (H,W) are the dimensions of the input image and N is n_bins or ``image.max() + 1`` if no value is provided as a parameter. Effectively, each pixel is a N-D feature vector that is the histogram. The sum of the elements in the feature vector will be 1, unless no pixels in the window were covered by both footprint and mask, in which case all elements will be 0. rI)rE) rOr:r;r<r=rJrKr>r?rBrCrCrD_apply_vector_per_pixels& rSselemr;z1.0z0.19) kwarg_mappingremoved_versiondeprecated_versionFc CFt|}|jdkrttj||||||dSttj|||||||dS)aBAuto-level image using local histogram. This filter locally stretches the histogram of gray values to cover the entire range of values from "white" to "black". Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. Examples -------- >>> from skimage import data >>> from skimage.morphology import disk, ball >>> from skimage.filters.rank import autolevel >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> auto = autolevel(img, disk(5)) >>> auto_vol = autolevel(volume, ball(5)) rr<r=rJrKr<r=rJrKrQ)r, asanyarrayr0rPr _autolevelrR _autolevel_3Dr:r;r<r=rJrKrQnp_imagerCrCrDr Ps +  r c CrX)aEqualize image using local histogram. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. Examples -------- >>> from skimage import data >>> from skimage.morphology import disk, ball >>> from skimage.filters.rank import equalize >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> equ = equalize(img, disk(5)) >>> equ_vol = equalize(volume, ball(5)) rrYrZ)r,r[r0rPr _equalizerR _equalize_3Dr^rCrCrDr  (  r c CrX)aReturn local gradient of an image (i.e. local maximum - local minimum). Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. Examples -------- >>> from skimage import data >>> from skimage.morphology import disk, ball >>> from skimage.filters.rank import gradient >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> out = gradient(img, disk(5)) >>> out_vol = gradient(volume, ball(5)) rrYrZ)r,r[r0rPr _gradientrR _gradient_3Dr^rCrCrDr rbr c CrX)aReturn local maximum of an image. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. See also -------- skimage.morphology.dilation Notes ----- The lower algorithm complexity makes `skimage.filters.rank.maximum` more efficient for larger images and footprints. Examples -------- >>> from skimage import data >>> from skimage.morphology import disk, ball >>> from skimage.filters.rank import maximum >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> out = maximum(img, disk(5)) >>> out_vol = maximum(volume, ball(5)) rrYrZ)r,r[r0rPr _maximumrR _maximum_3Dr^rCrCrDr  1  r c CrX)aReturn local mean of an image. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. Examples -------- >>> from skimage import data >>> from skimage.morphology import disk, ball >>> from skimage.filters.rank import mean >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> avg = mean(img, disk(5)) >>> avg_vol = mean(volume, ball(5)) rrYrZ)r,r[r0rPr _meanrR_mean_3Dr^rCrCrDr,rbrc CrX)a]Return local geometric mean of an image. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. Examples -------- >>> from skimage import data >>> from skimage.morphology import disk, ball >>> from skimage.filters.rank import mean >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> avg = geometric_mean(img, disk(5)) >>> avg_vol = geometric_mean(volume, ball(5)) References ---------- .. [1] Gonzalez, R. C. and Wood, R. E. "Digital Image Processing (3rd Edition)." Prentice-Hall Inc, 2006. rrYrZ)r,r[r0rPr _geometric_meanrR_geometric_mean_3Dr^rCrCrDr`s - rc CrX)aReturn image subtracted from its local mean. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. Notes ----- Subtracting the mean value may introduce underflow. To compensate this potential underflow, the obtained difference is downscaled by a factor of 2 and shifted by `n_bins / 2 - 1`, the median value of the local histogram (`n_bins = max(3, image.max()) +1` for 16-bits images and 256 otherwise). Examples -------- >>> from skimage import data >>> from skimage.morphology import disk, ball >>> from skimage.filters.rank import subtract_mean >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> out = subtract_mean(img, disk(5)) >>> out_vol = subtract_mean(volume, ball(5)) rrYrZ)r,r[r0rPr _subtract_meanrR_subtract_mean_3Dr^rCrCrDrs 0 rc Cs^t|}|durt|j|j}|jdkr"ttj||||||dSttj |||||||dS)atReturn local median of an image. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. If None, a full square of size 3 is used. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. See also -------- skimage.filters.median : Implementation of a median filtering which handles images with floating precision. Examples -------- >>> from skimage import data >>> from skimage.morphology import disk, ball >>> from skimage.filters.rank import median >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> med = median(img, disk(5)) >>> med_vol = median(volume, ball(5)) NrrYrZ) r,r[ndigenerate_binary_structurer0rPr _medianrR _median_3Dr^rCrCrDrs .  rc CrX)aReturn local minimum of an image. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. See also -------- skimage.morphology.erosion Notes ----- The lower algorithm complexity makes `skimage.filters.rank.minimum` more efficient for larger images and footprints. Examples -------- >>> from skimage import data >>> from skimage.morphology import disk, ball >>> from skimage.filters.rank import minimum >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> out = minimum(img, disk(5)) >>> out_vol = minimum(volume, ball(5)) rrYrZ)r,r[r0rPr _minimumrR _minimum_3Dr^rCrCrDrrgrc CrX)aReturn local mode of an image. The mode is the value that appears most often in the local histogram. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. Examples -------- >>> from skimage import data >>> from skimage.morphology import disk, ball >>> from skimage.filters.rank import modal >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> out = modal(img, disk(5)) >>> out_vol = modal(volume, ball(5)) rrYrZ)r,r[r0rPr _modalrR _modal_3Dr^rCrCrDrNs *  rc CrX)aEnhance contrast of an image. This replaces each pixel by the local maximum if the pixel gray value is closer to the local maximum than the local minimum. Otherwise it is replaced by the local minimum. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image Examples -------- >>> from skimage import data >>> from skimage.morphology import disk, ball >>> from skimage.filters.rank import enhance_contrast >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> out = enhance_contrast(img, disk(5)) >>> out_vol = enhance_contrast(volume, ball(5)) rrYrZ)r,r[r0rPr _enhance_contrastrR_enhance_contrast_3Dr^rCrCrDrs , rc CrX)aReturn the local number (population) of pixels. The number of pixels is defined as the number of pixels which are included in the footprint and the mask. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. Examples -------- >>> from skimage.morphology import square, cube # Need to add 3D example >>> import skimage.filters.rank as rank >>> img = 255 * np.array([[0, 0, 0, 0, 0], ... [0, 1, 1, 1, 0], ... [0, 1, 1, 1, 0], ... [0, 1, 1, 1, 0], ... [0, 0, 0, 0, 0]], dtype=np.uint8) >>> rank.pop(img, square(3)) array([[4, 6, 6, 6, 4], [6, 9, 9, 9, 6], [6, 9, 9, 9, 6], [6, 9, 9, 9, 6], [4, 6, 6, 6, 4]], dtype=uint8) rrYrZ)r,r[r0rPr _poprR_pop_3Dr^rCrCrDr /  rc CrX)aReturn the local sum of pixels. Note that the sum may overflow depending on the data type of the input array. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. Examples -------- >>> from skimage.morphology import square, cube # Need to add 3D example >>> import skimage.filters.rank as rank # Cube seems to fail but >>> img = np.array([[0, 0, 0, 0, 0], # Ball can pass ... [0, 1, 1, 1, 0], ... [0, 1, 1, 1, 0], ... [0, 1, 1, 1, 0], ... [0, 0, 0, 0, 0]], dtype=np.uint8) >>> rank.sum(img, square(3)) array([[1, 2, 3, 2, 1], [2, 4, 6, 4, 2], [3, 6, 9, 6, 3], [2, 4, 6, 4, 2], [1, 2, 3, 2, 1]], dtype=uint8) rrYrZ)r,r[r0rPr _sumrR_sum_3Dr^rCrCrDsumrzr}c CrX)aLocal threshold of an image. The resulting binary mask is True if the gray value of the center pixel is greater than the local mean. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. Examples -------- >>> from skimage.morphology import square, cube # Need to add 3D example >>> from skimage.filters.rank import threshold >>> img = 255 * np.array([[0, 0, 0, 0, 0], ... [0, 1, 1, 1, 0], ... [0, 1, 1, 1, 0], ... [0, 1, 1, 1, 0], ... [0, 0, 0, 0, 0]], dtype=np.uint8) >>> threshold(img, square(3)) array([[0, 0, 0, 0, 0], [0, 1, 1, 1, 0], [0, 1, 0, 1, 0], [0, 1, 1, 1, 0], [0, 0, 0, 0, 0]], dtype=uint8) rrYrZ)r,r[r0rPr _thresholdrR _threshold_3Dr^rCrCrDr2rzrc Cst|}|jdkr6t|jdd|}t|jdd|} |} d| || f<ttj|| ||||dSt|jdd|}t|jdd|} t|jdd|} |} d| || | f<t tj || |||||dS)a\Noise feature. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). References ---------- .. [1] N. Hashimoto et al. Referenceless image quality evaluation for whole slide imaging. J Pathol Inform 2012;3:9. Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. Examples -------- >>> from skimage import data >>> from skimage.morphology import disk, ball >>> from skimage.filters.rank import noise_filter >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> out = noise_filter(img, disk(5)) >>> out_vol = noise_filter(volume, ball(5)) rrrrYrZ) r,r[r0r7r3copyrPr _noise_filterrR_noise_filter_3D) r:r;r<r=rJrKrQr_centre_rcentre_c footprint_cpycentre_zrCrCrDrms( -  rc CsNt|}|jdkrttj||||||tjdSttj|||||||tjd S)aLocal entropy. The entropy is computed using base 2 logarithm i.e. the filter returns the minimum number of bits needed to encode the local gray level distribution. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (float) Output image. References ---------- .. [1] `https://en.wikipedia.org/wiki/Entropy_(information_theory) `_ Examples -------- >>> from skimage import data >>> from skimage.filters.rank import entropy >>> from skimage.morphology import disk, ball >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> ent = entropy(img, disk(5)) >>> ent_vol = entropy(volume, ball(5)) r)r<r=rJrKr>)r<r=rJrKrQr>) r,r[r0rPr _entropydoublerR _entropy_3Dr^rCrCrDrs 0  rc CrX)avLocal Otsu's threshold value for each pixel. Parameters ---------- image : ([P,] M, N) ndarray (uint8, uint16) Input image. footprint : ndarray The neighborhood expressed as an ndarray of 1's and 0's. out : ([P,] M, N) array (same dtype as input) If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y, shift_z : int Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : ([P,] M, N) ndarray (same dtype as input image) Output image. References ---------- .. [1] https://en.wikipedia.org/wiki/Otsu's_method Examples -------- >>> from skimage import data >>> from skimage.filters.rank import otsu >>> from skimage.morphology import disk, ball >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> local_otsu = otsu(img, disk(5)) >>> thresh_image = img >= local_otsu >>> local_otsu_vol = otsu(volume, ball(5)) >>> thresh_image_vol = volume >= local_otsu_vol rrYrZ)r,r[r0rPr _otsurR_otsu_3Dr^rCrCrDrs .  rc Cs6|dur t|d}ttj||||||tj|d S)aNormalized sliding window histogram Parameters ---------- image : 2-D array (integer or float) Input image. footprint : 2-D array (integer or float) The neighborhood expressed as a 2-D array of 1's and 0's. out : 2-D array (integer or float), optional If None, a new array is allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y : int, optional Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). n_bins : int or None The number of histogram bins. Will default to ``image.max() + 1`` if None is passed. Returns ------- out : 3-D array (float) Array of dimensions (H,W,N), where (H,W) are the dimensions of the input image and N is n_bins or ``image.max() + 1`` if no value is provided as a parameter. Effectively, each pixel is a N-D feature vector that is the histogram. The sum of the elements in the feature vector will be 1, unless no pixels in the window were covered by both footprint and mask, in which case all elements will be 0. Examples -------- >>> from skimage import data >>> from skimage.filters.rank import windowed_histogram >>> from skimage.morphology import disk, ball >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> hist_img = windowed_histogram(img, disk(5)) Nr)r<r=rJrKr>r?)r7r8rSr _windowed_histr,r)r:r;r<r=rJrKrBrCrCrDwindowed_histogram,s/ rrZc CrX)aAssign to each pixel the most common value within its neighborhood. Parameters ---------- image : ndarray Image array (uint8, uint16 array). footprint : 2-D array (integer or float) The neighborhood expressed as a 2-D array of 1's and 0's. out : ndarray (integer or float), optional If None, a new array will be allocated. mask : ndarray (integer or float), optional Mask array that defines (>0) area of the image included in the local neighborhood. If None, the complete image is used (default). shift_x, shift_y : int, optional Offset added to the footprint center point. Shift is bounded to the footprint sizes (center must be inside the given footprint). Returns ------- out : 2-D array (same dtype as input image) Output image. Examples -------- >>> from skimage import data >>> from skimage.filters.rank import majority >>> from skimage.morphology import disk, ball >>> import numpy as np >>> img = data.camera() >>> rng = np.random.default_rng() >>> volume = rng.integers(0, 255, size=(10,10,10), dtype=np.uint8) >>> maj_img = majority(img, disk(5)) >>> maj_img_vol = majority(volume, ball(5)) rrYrZ)r,r[r0rPr _majorityrR _majority_3Dr^rCrCrDmajorityes ( r)NNNNr)N)Nr)NNFFF)NNNFFF)NNFFN)'__doc__numpyr,scipyrrn _shared.utilsrrrutilrr __all__rErHrPrRrSr r r r rrrrrrrrr}rrrrrrrCrCrCrDs 2    V V &  2 5 2 2 ; 2 7 : : ; 4 6 9 9 9 F ; 8 7