1"""Standard test images. 2 3For more images, see 4 5 - http://sipi.usc.edu/database/database.php 6 7""" 8import numpy as np 9import shutil 10from packaging import version 11 12from ..util.dtype import img_as_bool 13from ._binary_blobs import binary_blobs 14from ._registry import registry, legacy_registry, registry_urls 15 16from .. import __version__ 17 18import os.path as osp 19import os 20 21legacy_data_dir = osp.abspath(osp.dirname(__file__)) 22skimage_distribution_dir = osp.join(legacy_data_dir, '..') 23 24try: 25 from pooch import file_hash 26except ModuleNotFoundError: 27 # Function taken from 28 # https://github.com/fatiando/pooch/blob/master/pooch/utils.py 29 def file_hash(fname, alg="sha256"): 30 """ 31 Calculate the hash of a given file. 32 Useful for checking if a file has changed or been corrupted. 33 Parameters 34 ---------- 35 fname : str 36 The name of the file. 37 alg : str 38 The type of the hashing algorithm 39 Returns 40 ------- 41 hash : str 42 The hash of the file. 43 Examples 44 -------- 45 >>> fname = "test-file-for-hash.txt" 46 >>> with open(fname, "w") as f: 47 ... __ = f.write("content of the file") 48 >>> print(file_hash(fname)) 49 0fc74468e6a9a829f103d069aeb2bb4f8646bad58bf146bb0e3379b759ec4a00 50 >>> import os 51 >>> os.remove(fname) 52 """ 53 import hashlib 54 if alg not in hashlib.algorithms_available: 55 raise ValueError(f'Algorithm \'{alg}\' not available in hashlib') 56 # Calculate the hash in chunks to avoid overloading the memory 57 chunksize = 65536 58 hasher = hashlib.new(alg) 59 with open(fname, "rb") as fin: 60 buff = fin.read(chunksize) 61 while buff: 62 hasher.update(buff) 63 buff = fin.read(chunksize) 64 return hasher.hexdigest() 65 66 67def _has_hash(path, expected_hash): 68 """Check if the provided path has the expected hash.""" 69 if not osp.exists(path): 70 return False 71 return file_hash(path) == expected_hash 72 73 74def create_image_fetcher(): 75 try: 76 import pooch 77 # older versions of Pooch don't have a __version__ attribute 78 if not hasattr(pooch, '__version__'): 79 retry = {} 80 else: 81 pooch_version = pooch.__version__.lstrip('v') 82 retry = {'retry_if_failed': 3} 83 # Keep version check in synch with 84 # scikit-image/requirements/optional.txt 85 if version.parse(pooch_version) < version.parse('1.3.0'): 86 # we need a more recent version of pooch to retry 87 retry = {} 88 except ImportError: 89 # Without pooch, fallback on the standard data directory 90 # which for now, includes a few limited data samples 91 return None, legacy_data_dir 92 93 # Pooch expects a `+` to exist in development versions. 94 # Since scikit-image doesn't follow that convention, we have to manually 95 # remove `.dev` with a `+` if it exists. 96 # This helps pooch understand that it should look in master 97 # to find the required files 98 if '+git' in __version__: 99 skimage_version_for_pooch = __version__.replace('.dev0+git', '+git') 100 else: 101 skimage_version_for_pooch = __version__.replace('.dev', '+') 102 103 if '+' in skimage_version_for_pooch: 104 url = ("https://github.com/scikit-image/scikit-image/raw/" 105 "{version}/skimage/") 106 else: 107 url = ("https://github.com/scikit-image/scikit-image/raw/" 108 "v{version}/skimage/") 109 110 # Create a new friend to manage your sample data storage 111 image_fetcher = pooch.create( 112 # Pooch uses appdirs to select an appropriate directory for the cache 113 # on each platform. 114 # https://github.com/ActiveState/appdirs 115 # On linux this converges to 116 # '$HOME/.cache/scikit-image' 117 # With a version qualifier 118 path=pooch.os_cache("scikit-image"), 119 base_url=url, 120 version=skimage_version_for_pooch, 121 version_dev="v0.19.x", 122 env="SKIMAGE_DATADIR", 123 registry=registry, 124 urls=registry_urls, 125 # Note: this should read `retry_if_failed=3,`, but we generate that 126 # dynamically at import time above, in case installed pooch is a less 127 # recent version 128 **retry, 129 ) 130 131 data_dir = osp.join(str(image_fetcher.abspath), 'data') 132 return image_fetcher, data_dir 133 134 135image_fetcher, data_dir = create_image_fetcher() 136 137if image_fetcher is None: 138 has_pooch = False 139else: 140 has_pooch = True 141 142 143def _fetch(data_filename): 144 """Fetch a given data file from either the local cache or the repository. 145 146 This function provides the path location of the data file given 147 its name in the scikit-image repository. 148 149 Parameters 150 ---------- 151 data_filename: 152 Name of the file in the scikit-image repository. e.g. 153 'restoration/tess/camera_rl.npy'. 154 155 Returns 156 ------- 157 Path of the local file as a python string. 158 159 Raises 160 ------ 161 KeyError: 162 If the filename is not known to the scikit-image distribution. 163 164 ModuleNotFoundError: 165 If the filename is known to the scikit-image distribution but pooch 166 is not installed. 167 168 ConnectionError: 169 If scikit-image is unable to connect to the internet but the 170 dataset has not been downloaded yet. 171 """ 172 resolved_path = osp.join(data_dir, '..', data_filename) 173 expected_hash = registry[data_filename] 174 175 # Case 1: 176 # The file may already be in the data_dir. 177 # We may have decided to ship it in the scikit-image distribution. 178 if _has_hash(resolved_path, expected_hash): 179 # Nothing to be done, file is where it is expected to be 180 return resolved_path 181 182 # Case 2: 183 # The user is using a cloned version of the github repo, which 184 # contains both the publicly shipped data, and test data. 185 # In this case, the file would be located relative to the 186 # skimage_distribution_dir 187 gh_repository_path = osp.join(skimage_distribution_dir, data_filename) 188 if _has_hash(gh_repository_path, expected_hash): 189 parent = osp.dirname(resolved_path) 190 os.makedirs(parent, exist_ok=True) 191 shutil.copy2(gh_repository_path, resolved_path) 192 return resolved_path 193 194 # Case 3: 195 # Pooch not found. 196 if image_fetcher is None: 197 raise ModuleNotFoundError( 198 "The requested file is part of the scikit-image distribution, " 199 "but requires the installation of an optional dependency, pooch. " 200 "To install pooch, use your preferred python package manager. " 201 "Follow installation instruction found at " 202 "https://scikit-image.org/docs/stable/install.html" 203 ) 204 205 # Case 4: 206 # Pooch needs to download the data. Let the image fetcher to search for 207 # our data. A ConnectionError is raised if no internet connection is 208 # available. 209 try: 210 resolved_path = image_fetcher.fetch(data_filename) 211 except ConnectionError as err: 212 # If we decide in the future to suppress the underlying 'requests' 213 # error, change this to `raise ... from None`. See PEP 3134. 214 raise ConnectionError( 215 'Tried to download a scikit-image dataset, but no internet ' 216 'connection is available. To avoid this message in the ' 217 'future, try `skimage.data.download_all()` when you are ' 218 'connected to the internet.' 219 ) from err 220 return resolved_path 221 222 223def _init_pooch(): 224 os.makedirs(data_dir, exist_ok=True) 225 226 # Copy in the README.txt if it doesn't already exist. 227 # If the file was originally copied to the data cache directory read-only 228 # then we cannot overwrite it, nor do we need to copy on every init. 229 # In general, as the data cache directory contains the scikit-image version 230 # it should not be necessary to overwrite this file as it should not 231 # change. 232 dest_path = osp.join(data_dir, 'README.txt') 233 if not os.path.isfile(dest_path): 234 shutil.copy2(osp.join(skimage_distribution_dir, 'data', 'README.txt'), 235 dest_path) 236 237 # Fetch all legacy data so that it is available by default 238 for filename in legacy_registry: 239 _fetch(filename) 240 241 242# This function creates directories, and has been the source of issues for 243# downstream users, see 244# https://github.com/scikit-image/scikit-image/issues/4660 245# https://github.com/scikit-image/scikit-image/issues/4664 246if has_pooch: 247 _init_pooch() 248 249 250def download_all(directory=None): 251 """Download all datasets for use with scikit-image offline. 252 253 Scikit-image datasets are no longer shipped with the library by default. 254 This allows us to use higher quality datasets, while keeping the 255 library download size small. 256 257 This function requires the installation of an optional dependency, pooch, 258 to download the full dataset. Follow installation instruction found at 259 260 https://scikit-image.org/docs/stable/install.html 261 262 Call this function to download all sample images making them available 263 offline on your machine. 264 265 Parameters 266 ---------- 267 directory: path-like, optional 268 The directory where the dataset should be stored. 269 270 Raises 271 ------ 272 ModuleNotFoundError: 273 If pooch is not install, this error will be raised. 274 275 Notes 276 ----- 277 scikit-image will only search for images stored in the default directory. 278 Only specify the directory if you wish to download the images to your own 279 folder for a particular reason. You can access the location of the default 280 data directory by inspecting the variable `skimage.data.data_dir`. 281 """ 282 283 if image_fetcher is None: 284 raise ModuleNotFoundError( 285 "To download all package data, scikit-image needs an optional " 286 "dependency, pooch." 287 "To install pooch, follow our installation instructions found at " 288 "https://scikit-image.org/docs/stable/install.html" 289 ) 290 # Consider moving this kind of logic to Pooch 291 old_dir = image_fetcher.path 292 try: 293 if directory is not None: 294 image_fetcher.path = directory 295 296 for filename in image_fetcher.registry: 297 _fetch(filename) 298 finally: 299 image_fetcher.path = old_dir 300 301 302def lbp_frontal_face_cascade_filename(): 303 """Return the path to the XML file containing the weak classifier cascade. 304 305 These classifiers were trained using LBP features. The file is part 306 of the OpenCV repository [1]_. 307 308 References 309 ---------- 310 .. [1] OpenCV lbpcascade trained files 311 https://github.com/opencv/opencv/tree/master/data/lbpcascades 312 """ 313 314 return _fetch('data/lbpcascade_frontalface_opencv.xml') 315 316 317def _load(f, as_gray=False): 318 """Load an image file located in the data directory. 319 320 Parameters 321 ---------- 322 f : string 323 File name. 324 as_gray : bool, optional 325 Whether to convert the image to grayscale. 326 327 Returns 328 ------- 329 img : ndarray 330 Image loaded from ``skimage.data_dir``. 331 """ 332 # importing io is quite slow since it scans all the backends 333 # we lazy import it here 334 from ..io import imread 335 return imread(_fetch(f), as_gray=as_gray) 336 337 338def camera(): 339 """Gray-level "camera" image. 340 341 Can be used for segmentation and denoising examples. 342 343 Returns 344 ------- 345 camera : (512, 512) uint8 ndarray 346 Camera image. 347 348 Notes 349 ----- 350 No copyright restrictions. CC0 by the photographer (Lav Varshney). 351 352 .. versionchanged:: 0.18 353 This image was replaced due to copyright restrictions. For more 354 information, please see [1]_. 355 356 References 357 ---------- 358 .. [1] https://github.com/scikit-image/scikit-image/issues/3927 359 """ 360 return _load("data/camera.png") 361 362 363def eagle(): 364 """A golden eagle. 365 366 Suitable for examples on segmentation, Hough transforms, and corner 367 detection. 368 369 Notes 370 ----- 371 No copyright restrictions. CC0 by the photographer (Dayane Machado). 372 373 Returns 374 ------- 375 eagle : (2019, 1826) uint8 ndarray 376 Eagle image. 377 """ 378 return _load("data/eagle.png") 379 380 381def astronaut(): 382 """Color image of the astronaut Eileen Collins. 383 384 Photograph of Eileen Collins, an American astronaut. She was selected 385 as an astronaut in 1992 and first piloted the space shuttle STS-63 in 386 1995. She retired in 2006 after spending a total of 38 days, 8 hours 387 and 10 minutes in outer space. 388 389 This image was downloaded from the NASA Great Images database 390 <https://flic.kr/p/r9qvLn>`__. 391 392 No known copyright restrictions, released into the public domain. 393 394 Returns 395 ------- 396 astronaut : (512, 512, 3) uint8 ndarray 397 Astronaut image. 398 """ 399 400 return _load("data/astronaut.png") 401 402 403def brick(): 404 """Brick wall. 405 406 Returns 407 ------- 408 brick : (512, 512) uint8 image 409 A small section of a brick wall. 410 411 Notes 412 ----- 413 The original image was downloaded from 414 `CC0Textures <https://cc0textures.com/view.php?tex=Bricks25>`_ and licensed 415 under the Creative Commons CC0 License. 416 417 A perspective transform was then applied to the image, prior to 418 rotating it by 90 degrees, cropping and scaling it to obtain the final 419 image. 420 """ 421 422 """ 423 The following code was used to obtain the final image. 424 425 >>> import sys; print(sys.version) 426 >>> import platform; print(platform.platform()) 427 >>> import skimage; print(f'scikit-image version: {skimage.__version__}') 428 >>> import numpy; print(f'numpy version: {numpy.__version__}') 429 >>> import imageio; print(f'imageio version {imageio.__version__}') 430 3.7.3 | packaged by conda-forge | (default, Jul 1 2019, 21:52:21) 431 [GCC 7.3.0] 432 Linux-5.0.0-20-generic-x86_64-with-debian-buster-sid 433 scikit-image version: 0.16.dev0 434 numpy version: 1.16.4 435 imageio version 2.4.1 436 437 >>> import requests 438 >>> import zipfile 439 >>> url = 'https://cdn.struffelproductions.com/file/cc0textures/Bricks25/%5B2K%5DBricks25.zip' 440 >>> r = requests.get(url) 441 >>> with open('[2K]Bricks25.zip', 'bw') as f: 442 ... f.write(r.content) 443 >>> with zipfile.ZipFile('[2K]Bricks25.zip') as z: 444 ... z.extract('Bricks25_col.jpg') 445 446 >>> from numpy.linalg import inv 447 >>> from skimage.transform import rescale, warp, rotate 448 >>> from skimage.color import rgb2gray 449 >>> from imageio import imread, imwrite 450 >>> from skimage import img_as_ubyte 451 >>> import numpy as np 452 453 454 >>> # Obtained playing around with GIMP 2.10 with their perspective tool 455 >>> H = inv(np.asarray([[ 0.54764, -0.00219, 0], 456 ... [-0.12822, 0.54688, 0], 457 ... [-0.00022, 0, 1]])) 458 459 460 >>> brick_orig = imread('Bricks25_col.jpg') 461 >>> brick = warp(brick_orig, H) 462 >>> brick = rescale(brick[:1024, :1024], (0.5, 0.5, 1)) 463 >>> brick = rotate(brick, -90) 464 >>> imwrite('brick.png', img_as_ubyte(rgb2gray(brick))) 465 """ 466 return _load("data/brick.png", as_gray=True) 467 468 469def grass(): 470 """Grass. 471 472 Returns 473 ------- 474 grass : (512, 512) uint8 image 475 Some grass. 476 477 Notes 478 ----- 479 The original image was downloaded from 480 `DeviantArt <https://www.deviantart.com/linolafett/art/Grass-01-434853879>`__ 481 and licensed under the Creative Commons CC0 License. 482 483 The downloaded image was cropped to include a region of ``(512, 512)`` 484 pixels around the top left corner, converted to grayscale, then to uint8 485 prior to saving the result in PNG format. 486 487 """ 488 489 """ 490 The following code was used to obtain the final image. 491 492 >>> import sys; print(sys.version) 493 >>> import platform; print(platform.platform()) 494 >>> import skimage; print(f'scikit-image version: {skimage.__version__}') 495 >>> import numpy; print(f'numpy version: {numpy.__version__}') 496 >>> import imageio; print(f'imageio version {imageio.__version__}') 497 3.7.3 | packaged by conda-forge | (default, Jul 1 2019, 21:52:21) 498 [GCC 7.3.0] 499 Linux-5.0.0-20-generic-x86_64-with-debian-buster-sid 500 scikit-image version: 0.16.dev0 501 numpy version: 1.16.4 502 imageio version 2.4.1 503 504 >>> import requests 505 >>> import zipfile 506 >>> url = 'https://images-wixmp-ed30a86b8c4ca887773594c2.wixmp.com/f/a407467e-4ff0-49f1-923f-c9e388e84612/d76wfef-2878b78d-5dce-43f9-be36-26ec9bc0df3b.jpg?token=eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJzdWIiOiJ1cm46YXBwOjdlMGQxODg5ODIyNjQzNzNhNWYwZDQxNWVhMGQyNmUwIiwiaXNzIjoidXJuOmFwcDo3ZTBkMTg4OTgyMjY0MzczYTVmMGQ0MTVlYTBkMjZlMCIsIm9iaiI6W1t7InBhdGgiOiJcL2ZcL2E0MDc0NjdlLTRmZjAtNDlmMS05MjNmLWM5ZTM4OGU4NDYxMlwvZDc2d2ZlZi0yODc4Yjc4ZC01ZGNlLTQzZjktYmUzNi0yNmVjOWJjMGRmM2IuanBnIn1dXSwiYXVkIjpbInVybjpzZXJ2aWNlOmZpbGUuZG93bmxvYWQiXX0.98hIcOTCqXWQ67Ec5bM5eovKEn2p91mWB3uedH61ynI' 507 >>> r = requests.get(url) 508 >>> with open('grass_orig.jpg', 'bw') as f: 509 ... f.write(r.content) 510 >>> grass_orig = imageio.imread('grass_orig.jpg') 511 >>> grass = skimage.img_as_ubyte(skimage.color.rgb2gray(grass_orig[:512, :512])) 512 >>> imageio.imwrite('grass.png', grass) 513 """ 514 return _load("data/grass.png", as_gray=True) 515 516 517def gravel(): 518 """Gravel 519 520 Returns 521 ------- 522 gravel : (512, 512) uint8 image 523 Grayscale gravel sample. 524 525 Notes 526 ----- 527 The original image was downloaded from 528 `CC0Textures <https://cc0textures.com/view.php?tex=Gravel04>`__ and 529 licensed under the Creative Commons CC0 License. 530 531 The downloaded image was then rescaled to ``(1024, 1024)``, then the 532 top left ``(512, 512)`` pixel region was cropped prior to converting the 533 image to grayscale and uint8 data type. The result was saved using the 534 PNG format. 535 """ 536 537 """ 538 The following code was used to obtain the final image. 539 540 >>> import sys; print(sys.version) 541 >>> import platform; print(platform.platform()) 542 >>> import skimage; print(f'scikit-image version: {skimage.__version__}') 543 >>> import numpy; print(f'numpy version: {numpy.__version__}') 544 >>> import imageio; print(f'imageio version {imageio.__version__}') 545 3.7.3 | packaged by conda-forge | (default, Jul 1 2019, 21:52:21) 546 [GCC 7.3.0] 547 Linux-5.0.0-20-generic-x86_64-with-debian-buster-sid 548 scikit-image version: 0.16.dev0 549 numpy version: 1.16.4 550 imageio version 2.4.1 551 552 >>> import requests 553 >>> import zipfile 554 555 >>> url = 'https://cdn.struffelproductions.com/file/cc0textures/Gravel04/%5B2K%5DGravel04.zip' 556 >>> r = requests.get(url) 557 >>> with open('[2K]Gravel04.zip', 'bw') as f: 558 ... f.write(r.content) 559 560 >>> with zipfile.ZipFile('[2K]Gravel04.zip') as z: 561 ... z.extract('Gravel04_col.jpg') 562 563 >>> from skimage.transform import resize 564 >>> gravel_orig = imageio.imread('Gravel04_col.jpg') 565 >>> gravel = resize(gravel_orig, (1024, 1024)) 566 >>> gravel = skimage.img_as_ubyte(skimage.color.rgb2gray(gravel[:512, :512])) 567 >>> imageio.imwrite('gravel.png', gravel) 568 """ 569 return _load("data/gravel.png", as_gray=True) 570 571 572def text(): 573 """Gray-level "text" image used for corner detection. 574 575 Notes 576 ----- 577 This image was downloaded from Wikipedia 578 <https://en.wikipedia.org/wiki/File:Corner.png>`__. 579 580 No known copyright restrictions, released into the public domain. 581 582 Returns 583 ------- 584 text : (172, 448) uint8 ndarray 585 Text image. 586 """ 587 588 return _load("data/text.png") 589 590 591def checkerboard(): 592 """Checkerboard image. 593 594 Checkerboards are often used in image calibration, since the 595 corner-points are easy to locate. Because of the many parallel 596 edges, they also visualise distortions particularly well. 597 598 Returns 599 ------- 600 checkerboard : (200, 200) uint8 ndarray 601 Checkerboard image. 602 """ 603 return _load("data/chessboard_GRAY.png") 604 605 606def cells3d(): 607 """3D fluorescence microscopy image of cells. 608 609 The returned data is a 3D multichannel array with dimensions provided in 610 ``(z, c, y, x)`` order. Each voxel has a size of ``(0.29 0.26 0.26)`` 611 micrometer. Channel 0 contains cell membranes, channel 1 contains nuclei. 612 613 Returns 614 ------- 615 cells3d: (60, 2, 256, 256) uint16 ndarray 616 The volumetric images of cells taken with an optical microscope. 617 618 Notes 619 ----- 620 The data for this was provided by the Allen Institute for Cell Science. 621 622 It has been downsampled by a factor of 4 in the row and column dimensions 623 to reduce computational time. 624 625 The microscope reports the following voxel spacing in microns: 626 627 * Original voxel size is ``(0.290, 0.065, 0.065)``. 628 * Scaling factor is ``(1, 4, 4)`` in each dimension. 629 * After rescaling the voxel size is ``(0.29 0.26 0.26)``. 630 """ 631 632 return _load("data/cells3d.tif") 633 634 635def human_mitosis(): 636 """Image of human cells undergoing mitosis. 637 638 Returns 639 ------- 640 human_mitosis: (512, 512) uint8 ndimage 641 Data of human cells undergoing mitosis taken during the preperation 642 of the manuscript in [1]_. 643 644 Notes 645 ----- 646 Copyright David Root. Licensed under CC-0 [2]_. 647 648 References 649 ---------- 650 .. [1] Moffat J, Grueneberg DA, Yang X, Kim SY, Kloepfer AM, Hinkle G, 651 Piqani B, Eisenhaure TM, Luo B, Grenier JK, Carpenter AE, Foo SY, 652 Stewart SA, Stockwell BR, Hacohen N, Hahn WC, Lander ES, 653 Sabatini DM, Root DE (2006) A lentiviral RNAi library for human and 654 mouse genes applied to an arrayed viral high-content screen. Cell, 655 124(6):1283-98 / :DOI: `10.1016/j.cell.2006.01.040` PMID 16564017 656 657 .. [2] GitHub licensing discussion 658 https://github.com/CellProfiler/examples/issues/41 659 660 """ 661 return _load('data/mitosis.tif') 662 663 664def cell(): 665 """Cell floating in saline. 666 667 This is a quantitative phase image retrieved from a digital hologram using 668 the Python library ``qpformat``. The image shows a cell with high phase 669 value, above the background phase. 670 671 Because of a banding pattern artifact in the background, this image is a 672 good test of thresholding algorithms. The pixel spacing is 0.107 µm. 673 674 These data were part of a comparison between several refractive index 675 retrieval techniques for spherical objects as part of [1]_. 676 677 This image is CC0, dedicated to the public domain. You may copy, modify, or 678 distribute it without asking permission. 679 680 Returns 681 ------- 682 cell : (660, 550) uint8 array 683 Image of a cell. 684 685 References 686 ---------- 687 .. [1] Paul Müller, Mirjam Schürmann, Salvatore Girardo, Gheorghe Cojoc, 688 and Jochen Guck. "Accurate evaluation of size and refractive index 689 for spherical objects in quantitative phase imaging." Optics Express 690 26(8): 10729-10743 (2018). :DOI:`10.1364/OE.26.010729` 691 """ 692 return _load('data/cell.png') 693 694 695def coins(): 696 """Greek coins from Pompeii. 697 698 This image shows several coins outlined against a gray background. 699 It is especially useful in, e.g. segmentation tests, where 700 individual objects need to be identified against a background. 701 The background shares enough grey levels with the coins that a 702 simple segmentation is not sufficient. 703 704 Notes 705 ----- 706 This image was downloaded from the 707 `Brooklyn Museum Collection 708 <https://www.brooklynmuseum.org/opencollection/archives/image/51611>`__. 709 710 No known copyright restrictions. 711 712 Returns 713 ------- 714 coins : (303, 384) uint8 ndarray 715 Coins image. 716 """ 717 return _load("data/coins.png") 718 719 720def kidney(): 721 """Mouse kidney tissue. 722 723 This biological tissue on a pre-prepared slide was imaged with confocal 724 fluorescence microscopy (Nikon C1 inverted microscope). 725 Image shape is (16, 512, 512, 3). That is 512x512 pixels in X-Y, 726 16 image slices in Z, and 3 color channels 727 (emission wavelengths 450nm, 515nm, and 605nm, respectively). 728 Real-space voxel size is 1.24 microns in X-Y, and 1.25 microns in Z. 729 Data type is unsigned 16-bit integers. 730 731 Notes 732 ----- 733 This image was acquired by Genevieve Buckley at Monasoh Micro Imaging in 734 2018. 735 License: CC0 736 737 Returns 738 ------- 739 kidney : (16, 512, 512, 3) uint16 ndarray 740 Kidney 3D multichannel image. 741 """ 742 return _load("data/kidney.tif") 743 744 745def lily(): 746 """Lily of the valley plant stem. 747 748 This plant stem on a pre-prepared slide was imaged with confocal 749 fluorescence microscopy (Nikon C1 inverted microscope). 750 Image shape is (922, 922, 4). That is 922x922 pixels in X-Y, 751 with 4 color channels. 752 Real-space voxel size is 1.24 microns in X-Y. 753 Data type is unsigned 16-bit integers. 754 755 Notes 756 ----- 757 This image was acquired by Genevieve Buckley at Monasoh Micro Imaging in 758 2018. 759 License: CC0 760 761 Returns 762 ------- 763 lily : (922, 922, 4) uint16 ndarray 764 Lily 2D multichannel image. 765 """ 766 return _load("data/lily.tif") 767 768 769def logo(): 770 """Scikit-image logo, a RGBA image. 771 772 Returns 773 ------- 774 logo : (500, 500, 4) uint8 ndarray 775 Logo image. 776 """ 777 return _load("data/logo.png") 778 779 780def microaneurysms(): 781 """Gray-level "microaneurysms" image. 782 783 Detail from an image of the retina (green channel). 784 The image is a crop of image 07_dr.JPG from the 785 High-Resolution Fundus (HRF) Image Database: 786 https://www5.cs.fau.de/research/data/fundus-images/ 787 788 Notes 789 ----- 790 No copyright restrictions. CC0 given by owner (Andreas Maier). 791 792 Returns 793 ------- 794 microaneurysms : (102, 102) uint8 ndarray 795 Retina image with lesions. 796 797 References 798 ---------- 799 .. [1] Budai, A., Bock, R, Maier, A., Hornegger, J., 800 Michelson, G. (2013). Robust Vessel Segmentation in Fundus 801 Images. International Journal of Biomedical Imaging, vol. 2013, 802 2013. 803 :DOI:`10.1155/2013/154860` 804 """ 805 return _load("data/microaneurysms.png") 806 807 808def moon(): 809 """Surface of the moon. 810 811 This low-contrast image of the surface of the moon is useful for 812 illustrating histogram equalization and contrast stretching. 813 814 Returns 815 ------- 816 moon : (512, 512) uint8 ndarray 817 Moon image. 818 """ 819 return _load("data/moon.png") 820 821 822def page(): 823 """Scanned page. 824 825 This image of printed text is useful for demonstrations requiring uneven 826 background illumination. 827 828 Returns 829 ------- 830 page : (191, 384) uint8 ndarray 831 Page image. 832 """ 833 return _load("data/page.png") 834 835 836def horse(): 837 """Black and white silhouette of a horse. 838 839 This image was downloaded from 840 `openclipart <http://openclipart.org/detail/158377/horse-by-marauder>` 841 842 No copyright restrictions. CC0 given by owner (Andreas Preuss (marauder)). 843 844 Returns 845 ------- 846 horse : (328, 400) bool ndarray 847 Horse image. 848 """ 849 return img_as_bool(_load("data/horse.png", as_gray=True)) 850 851 852def clock(): 853 """Motion blurred clock. 854 855 This photograph of a wall clock was taken while moving the camera in an 856 aproximately horizontal direction. It may be used to illustrate 857 inverse filters and deconvolution. 858 859 Released into the public domain by the photographer (Stefan van der Walt). 860 861 Returns 862 ------- 863 clock : (300, 400) uint8 ndarray 864 Clock image. 865 """ 866 return _load("data/clock_motion.png") 867 868 869def immunohistochemistry(): 870 """Immunohistochemical (IHC) staining with hematoxylin counterstaining. 871 872 This picture shows colonic glands where the IHC expression of FHL2 protein 873 is revealed with DAB. Hematoxylin counterstaining is applied to enhance the 874 negative parts of the tissue. 875 876 This image was acquired at the Center for Microscopy And Molecular Imaging 877 (CMMI). 878 879 No known copyright restrictions. 880 881 Returns 882 ------- 883 immunohistochemistry : (512, 512, 3) uint8 ndarray 884 Immunohistochemistry image. 885 """ 886 return _load("data/ihc.png") 887 888 889def chelsea(): 890 """Chelsea the cat. 891 892 An example with texture, prominent edges in horizontal and diagonal 893 directions, as well as features of differing scales. 894 895 Notes 896 ----- 897 No copyright restrictions. CC0 by the photographer (Stefan van der Walt). 898 899 Returns 900 ------- 901 chelsea : (300, 451, 3) uint8 ndarray 902 Chelsea image. 903 """ 904 return _load("data/chelsea.png") 905 906 907# Define an alias for chelsea that is more descriptive. 908cat = chelsea 909 910 911def coffee(): 912 """Coffee cup. 913 914 This photograph is courtesy of Pikolo Espresso Bar. 915 It contains several elliptical shapes as well as varying texture (smooth 916 porcelain to course wood grain). 917 918 Notes 919 ----- 920 No copyright restrictions. CC0 by the photographer (Rachel Michetti). 921 922 Returns 923 ------- 924 coffee : (400, 600, 3) uint8 ndarray 925 Coffee image. 926 """ 927 return _load("data/coffee.png") 928 929 930def hubble_deep_field(): 931 """Hubble eXtreme Deep Field. 932 933 This photograph contains the Hubble Telescope's farthest ever view of 934 the universe. It can be useful as an example for multi-scale 935 detection. 936 937 Notes 938 ----- 939 This image was downloaded from 940 `HubbleSite 941 <http://hubblesite.org/newscenter/archive/releases/2012/37/image/a/>`__. 942 943 The image was captured by NASA and `may be freely used in the public domain 944 <http://www.nasa.gov/audience/formedia/features/MP_Photo_Guidelines.html>`_. 945 946 Returns 947 ------- 948 hubble_deep_field : (872, 1000, 3) uint8 ndarray 949 Hubble deep field image. 950 """ 951 return _load("data/hubble_deep_field.jpg") 952 953 954def retina(): 955 """Human retina. 956 957 This image of a retina is useful for demonstrations requiring circular 958 images. 959 960 Notes 961 ----- 962 This image was downloaded from 963 `wikimedia <https://commons.wikimedia.org/wiki/File:Fundus_photograph_of_normal_left_eye.jpg>`. 964 This file is made available under the Creative Commons CC0 1.0 Universal 965 Public Domain Dedication. 966 967 References 968 ---------- 969 .. [1] Häggström, Mikael (2014). "Medical gallery of Mikael Häggström 2014". 970 WikiJournal of Medicine 1 (2). :DOI:`10.15347/wjm/2014.008`. 971 ISSN 2002-4436. Public Domain 972 973 Returns 974 ------- 975 retina : (1411, 1411, 3) uint8 ndarray 976 Retina image in RGB. 977 """ 978 return _load("data/retina.jpg") 979 980 981def shepp_logan_phantom(): 982 """Shepp Logan Phantom. 983 984 References 985 ---------- 986 .. [1] L. A. Shepp and B. F. Logan, "The Fourier reconstruction of a head 987 section," in IEEE Transactions on Nuclear Science, vol. 21, 988 no. 3, pp. 21-43, June 1974. :DOI:`10.1109/TNS.1974.6499235` 989 990 Returns 991 ------- 992 phantom : (400, 400) float64 image 993 Image of the Shepp-Logan phantom in grayscale. 994 """ 995 return _load("data/phantom.png", as_gray=True) 996 997 998def colorwheel(): 999 """Color Wheel. 1000 1001 Returns 1002 ------- 1003 colorwheel : (370, 371, 3) uint8 image 1004 A colorwheel. 1005 """ 1006 return _load("data/color.png") 1007 1008 1009def rocket(): 1010 """Launch photo of DSCOVR on Falcon 9 by SpaceX. 1011 1012 This is the launch photo of Falcon 9 carrying DSCOVR lifted off from 1013 SpaceX's Launch Complex 40 at Cape Canaveral Air Force Station, FL. 1014 1015 Notes 1016 ----- 1017 This image was downloaded from 1018 `SpaceX Photos 1019 <https://www.flickr.com/photos/spacexphotos/16511594820/in/photostream/>`__. 1020 1021 The image was captured by SpaceX and `released in the public domain 1022 <http://arstechnica.com/tech-policy/2015/03/elon-musk-puts-spacex-photos-into-the-public-domain/>`_. 1023 1024 Returns 1025 ------- 1026 rocket : (427, 640, 3) uint8 ndarray 1027 Rocket image. 1028 """ 1029 return _load("data/rocket.jpg") 1030 1031 1032def stereo_motorcycle(): 1033 """Rectified stereo image pair with ground-truth disparities. 1034 1035 The two images are rectified such that every pixel in the left image has 1036 its corresponding pixel on the same scanline in the right image. That means 1037 that both images are warped such that they have the same orientation but a 1038 horizontal spatial offset (baseline). The ground-truth pixel offset in 1039 column direction is specified by the included disparity map. 1040 1041 The two images are part of the Middlebury 2014 stereo benchmark. The 1042 dataset was created by Nera Nesic, Porter Westling, Xi Wang, York Kitajima, 1043 Greg Krathwohl, and Daniel Scharstein at Middlebury College. A detailed 1044 description of the acquisition process can be found in [1]_. 1045 1046 The images included here are down-sampled versions of the default exposure 1047 images in the benchmark. The images are down-sampled by a factor of 4 using 1048 the function `skimage.transform.downscale_local_mean`. The calibration data 1049 in the following and the included ground-truth disparity map are valid for 1050 the down-sampled images:: 1051 1052 Focal length: 994.978px 1053 Principal point x: 311.193px 1054 Principal point y: 254.877px 1055 Principal point dx: 31.086px 1056 Baseline: 193.001mm 1057 1058 Returns 1059 ------- 1060 img_left : (500, 741, 3) uint8 ndarray 1061 Left stereo image. 1062 img_right : (500, 741, 3) uint8 ndarray 1063 Right stereo image. 1064 disp : (500, 741, 3) float ndarray 1065 Ground-truth disparity map, where each value describes the offset in 1066 column direction between corresponding pixels in the left and the right 1067 stereo images. E.g. the corresponding pixel of 1068 ``img_left[10, 10 + disp[10, 10]]`` is ``img_right[10, 10]``. 1069 NaNs denote pixels in the left image that do not have ground-truth. 1070 1071 Notes 1072 ----- 1073 The original resolution images, images with different exposure and 1074 lighting, and ground-truth depth maps can be found at the Middlebury 1075 website [2]_. 1076 1077 References 1078 ---------- 1079 .. [1] D. Scharstein, H. Hirschmueller, Y. Kitajima, G. Krathwohl, N. 1080 Nesic, X. Wang, and P. Westling. High-resolution stereo datasets 1081 with subpixel-accurate ground truth. In German Conference on Pattern 1082 Recognition (GCPR 2014), Muenster, Germany, September 2014. 1083 .. [2] http://vision.middlebury.edu/stereo/data/scenes2014/ 1084 1085 """ 1086 filename = _fetch("data/motorcycle_disp.npz") 1087 disp = np.load(filename)['arr_0'] 1088 return (_load("data/motorcycle_left.png"), 1089 _load("data/motorcycle_right.png"), 1090 disp) 1091 1092 1093def lfw_subset(): 1094 """Subset of data from the LFW dataset. 1095 1096 This database is a subset of the LFW database containing: 1097 1098 * 100 faces 1099 * 100 non-faces 1100 1101 The full dataset is available at [2]_. 1102 1103 Returns 1104 ------- 1105 images : (200, 25, 25) uint8 ndarray 1106 100 first images are faces and subsequent 100 are non-faces. 1107 1108 Notes 1109 ----- 1110 The faces were randomly selected from the LFW dataset and the non-faces 1111 were extracted from the background of the same dataset. The cropped ROIs 1112 have been resized to a 25 x 25 pixels. 1113 1114 References 1115 ---------- 1116 .. [1] Huang, G., Mattar, M., Lee, H., & Learned-Miller, E. G. (2012). 1117 Learning to align from scratch. In Advances in Neural Information 1118 Processing Systems (pp. 764-772). 1119 .. [2] http://vis-www.cs.umass.edu/lfw/ 1120 1121 """ 1122 return np.load(_fetch('data/lfw_subset.npy')) 1123 1124 1125def skin(): 1126 """Microscopy image of dermis and epidermis (skin layers). 1127 1128 Hematoxylin and eosin stained slide at 10x of normal epidermis and dermis 1129 with a benign intradermal nevus. 1130 1131 Notes 1132 ----- 1133 This image requires an Internet connection the first time it is called, 1134 and to have the ``pooch`` package installed, in order to fetch the image 1135 file from the scikit-image datasets repository. 1136 1137 The source of this image is 1138 https://en.wikipedia.org/wiki/File:Normal_Epidermis_and_Dermis_with_Intradermal_Nevus_10x.JPG 1139 1140 The image was released in the public domain by its author Kilbad. 1141 1142 Returns 1143 ------- 1144 skin : (960, 1280, 3) RGB image of uint8 1145 """ 1146 return _load('data/skin.jpg') 1147 1148 1149def brain(): 1150 """Subset of data from the University of North Carolina Volume Rendering 1151 Test Data Set. 1152 1153 The full dataset is available at [1]_. 1154 1155 Returns 1156 ------- 1157 image : (10, 256, 256) uint16 ndarray 1158 1159 Notes 1160 ----- 1161 The 3D volume consists of 10 layers from the larger volume. 1162 1163 References 1164 ---------- 1165 .. [1] https://graphics.stanford.edu/data/voldata/ 1166 1167 """ 1168 return _load("data/brain.tiff") 1169 1170 1171def vortex(): 1172 """Case B1 image pair from the first PIV challenge. 1173 1174 Returns 1175 ------- 1176 image0, image1 : (512, 512) grayscale images 1177 A pair of images featuring synthetic moving particles. 1178 1179 Notes 1180 ----- 1181 This image was licensed as CC0 by its author, Prof. Koji Okamoto, with 1182 thanks to Prof. Jun Sakakibara, who maintains the PIV Challenge site. 1183 1184 References 1185 ---------- 1186 .. [1] Particle Image Velocimetry (PIV) Challenge site 1187 http://pivchallenge.org 1188 .. [2] 1st PIV challenge Case B: http://pivchallenge.org/pub/index.html#b 1189 """ 1190 return (_load('data/pivchallenge-B-B001_1.tif'), 1191 _load('data/pivchallenge-B-B001_2.tif')) 1192
