peak_local_max attribute 'min_distance' issue #5598

taroko-mooncake · 2021-10-11T17:01:12Z

Description

The attribute 'min_distance' in peak_local_max function is not working. Distance can be set at any integer and coordinates of centres remain unchanged.

Way to reproduce

distance = ndi.distance_transform_edt(binary_mask)
coords = peak_local_max(distance, footprint=np.ones((50, 50)), labels=binary_mask, threshold_rel=0.3)

Version information

# Paste the output of the following python commands
3.7.10 (default, Jun  4 2021, 14:48:32) 
[GCC 7.5.0]
Linux-4.14.243-185.433.amzn2.x86_64-x86_64-with-debian-10.6
scikit-image version: 0.18.1
numpy version: 1.20.3

/opt/conda/lib/python3.7/site-packages/matplotlib/image.py:446: RuntimeWarning: overflow encountered in double_scalars
  newmin = vmid - dv * fact
/opt/conda/lib/python3.7/site-packages/matplotlib/image.py:451: RuntimeWarning: overflow encountered in double_scalars
  newmax = vmid + dv * fact
/opt/conda/lib/python3.7/site-packages/matplotlib/image.py:503: RuntimeWarning: invalid value encountered in multiply
  A_resampled *= ((a_max - a_min) / frac)
/opt/conda/lib/python3.7/site-packages/matplotlib/image.py:504: RuntimeWarning: invalid value encountered in multiply
  vrange *= ((a_max - a_min) / frac)

alexdesiqueira · 2021-10-11T18:25:08Z

Thanks for opening this issue @taroko-mooncake! Could you please provide us a full, runnable example for us to check this bug?
Thanks again!

taroko-mooncake · 2021-10-12T08:33:53Z

# Generate an initial image with two overlapping circles
x, y = np.indices((80, 80))
x1, y1, x2, y2 = 28, 28, 44, 52
r1, r2 = 16, 20
mask_circle1 = (x - x1)**2 + (y - y1)**2 < r1**2
mask_circle2 = (x - x2)**2 + (y - y2)**2 < r2**2
image = np.logical_or(mask_circle1, mask_circle2)

# Now we want to separate the two objects in image
# Generate the markers as local maxima of the distance to the background
distance = ndi.distance_transform_edt(image)
coords = peak_local_max(distance, footprint=np.ones((3, 3)), labels=image)`

rfezzani · 2021-10-12T08:55:44Z

@taroko-mooncake, can you please be more precise in your bug report. Displaying the result of your provided example gives

fig, (ax0, ax1) = plt.subplots(1, 2)
ax1.imshow(distance)
ax0.imshow(image)
ax1.plot(coords[:, 1], coords[:, 0], 'rx')

Isn't it what you expected?

taroko-mooncake · 2021-10-12T12:14:45Z

Yes the algorithm works but if you apply 'min_distance' attribute, nothing changes.

# Generate an initial image with two overlapping circles
x, y = np.indices((80, 80))
x1, y1, x2, y2 = 28, 28, 44, 52
r1, r2 = 16, 20
mask_circle1 = (x - x1)**2 + (y - y1)**2 < r1**2
mask_circle2 = (x - x2)**2 + (y - y2)**2 < r2**2
image = np.logical_or(mask_circle1, mask_circle2)

# Now we want to separate the two objects in image
# Generate the markers as local maxima of the distance to the background
distance = ndi.distance_transform_edt(image)
coords = peak_local_max(distance, footprint=np.ones((3, 3)), labels=image, min_distance=70)`

There is something about min_distance that is causing a bug but I can't tell you anymore than that.

rfezzani · 2021-10-22T09:41:36Z

@taroko-mooncake, as the Note section stipulates in the doc:

The peak local maximum function returns the coordinates of local peaks (maxima) in an image. Internally, a maximum filter is used for finding local maxima. This operation dilates the original image. After comparison of the dilated and original image, this function returns the coordinates or a mask of the peaks where the dilated image equals the original image.

So the peak candidates are selected from this image

obtained using

import numpy as np
from scipy import ndimage as ndi
import matplotlib.pyplot as plt

from skimage.feature import peak_local_max

# Generate an initial image with two overlapping circles
x, y = np.indices((80, 80))
x1, y1, x2, y2 = 28, 28, 44, 52
r1, r2 = 16, 20
mask_circle1 = (x - x1)**2 + (y - y1)**2 < r1**2
mask_circle2 = (x - x2)**2 + (y - y2)**2 < r2**2
image = np.logical_or(mask_circle1, mask_circle2)

# Now we want to separate the two objects in image
# Generate the markers as local maxima of the distance to the background
distance = ndi.distance_transform_edt(image)
footprint = np.ones((3, 3))

# Show peak candidates
plt.imshow(ndi.maximum_filter(distance, footprint=footprint,
                              mode='constant') == distance)

Now with different values for min_distance, we obtain

# Show the output with different min_distance
fig, ax_list = plt.subplots(2, 2, figsize=(5, 5))
for ax, min_dist in zip(ax_list.ravel(), [1, 5, 25, 30]):
    coords = peak_local_max(distance, footprint=np.ones((3, 3)),
                            min_distance=min_dist)

    ax.imshow(distance)
    ax.plot(coords[:, 1], coords[:, 0], 'rx')
    ax.set_title(f"min_distance={min_dist}")
fig.tight_layout()
plt.show()

Please note that with min_distance=30 no peaks are detected because exclude_border is True by default and pixels within min_distance from the border are excluded.

jonaszeuner · 2021-12-30T10:24:15Z

Hi,
I also ran into an issue with the min_distance feature when using experimental data. My understanding of min_distance is, that the peak coordinates are not allowed to be closer than the value specified as min_distance. In the code example below this is clearly not the case (min_distance=10).

The runnable code including the image as float array below.

img=[[0.07185473, 0.07237354, 0.07473869, 0.07588312, 0.07338064,
        0.0685893 , 0.06788739, 0.07005417, 0.07190051, 0.07367056,
        0.07673762, 0.08041505, 0.08322271, 0.09103532, 0.0942855 ,
        0.09536889, 0.0936141 , 0.09129473, 0.09097429, 0.08816663,
        0.09777981, 0.10525673, 0.10922408, 0.11287098, 0.11616693,
        0.12251469, 0.12773327, 0.12834363, 0.13076982, 0.13530175],
       [0.07287709, 0.07579156, 0.07939269, 0.07614252, 0.07319753,
        0.07020676, 0.06991684, 0.06982528, 0.06932174, 0.07158007,
        0.07747005, 0.08441291, 0.09262226, 0.10464637, 0.11248951,
        0.11317617, 0.11308461, 0.11239796, 0.10971237, 0.11168078,
        0.12133974, 0.12573434, 0.12942702, 0.1328603 , 0.1363241 ,
        0.14705119, 0.15480278, 0.15773251, 0.15944152, 0.15269703],
       [0.07690547, 0.0801709 , 0.07913329, 0.07707332, 0.07710384,
        0.07612726, 0.07466239, 0.07159533, 0.07254139, 0.07891966,
        0.08923476, 0.1105211 , 0.13031205, 0.14895857, 0.16563668,
        0.1654078 , 0.15901427, 0.14918746, 0.13527123, 0.13517967,
        0.14599832, 0.15809873, 0.17221332, 0.17415122, 0.178851  ,
        0.17650111, 0.18323033, 0.18271153, 0.18051423, 0.17975128],
       [0.08044556, 0.08233768, 0.08540475, 0.08058289, 0.08056764,
        0.07867552, 0.0754406 , 0.07289235, 0.07115282, 0.08264286,
        0.1057145 , 0.13731594, 0.17752346, 0.21478599, 0.24649424,
        0.24957656, 0.24103151, 0.21493858, 0.20553902, 0.19511711,
        0.19822995, 0.20770581, 0.22546731, 0.23237964, 0.23430228,
        0.23083848, 0.21844816, 0.22291905, 0.22032502, 0.20735485],
       [0.07713436, 0.07962158, 0.08088808, 0.07916381, 0.07756161,
        0.07371633, 0.07545586, 0.0736553 , 0.08253605, 0.10061799,
        0.13951324, 0.19591058, 0.2620737 , 0.32628367, 0.35275807,
        0.35532158, 0.34709697, 0.32512398, 0.2867628 , 0.25554284,
        0.25499351, 0.26767376, 0.2804303 , 0.28729686, 0.28651865,
        0.27038987, 0.25963226, 0.24295415, 0.24213016, 0.23714046],
       [0.07457084, 0.07481498, 0.07519646, 0.07612726, 0.07734798,
        0.07537957, 0.07676814, 0.08091859, 0.09613184, 0.13318074,
        0.19253834, 0.28331426, 0.36543832, 0.44122988, 0.48641184,
        0.49250019, 0.47475395, 0.43721675, 0.38077363, 0.36330205,
        0.35648127, 0.35898375, 0.36124208, 0.35980774, 0.35187304,
        0.32892348, 0.30913252, 0.28918898, 0.28278019, 0.26912337],
       [0.07692073, 0.07153429, 0.07174792, 0.07307546, 0.07586786,
        0.07777523, 0.0816968 , 0.0914168 , 0.11578546, 0.16229496,
        0.24623484, 0.36723888, 0.49283589, 0.6037995 , 0.66649882,
        0.68049134, 0.65001907, 0.5857023 , 0.53644617, 0.48502327,
        0.45104143, 0.44589914, 0.43721675, 0.41820401, 0.40958267,
        0.39359121, 0.35474174, 0.31502251, 0.30200656, 0.29286641],
       [0.07362478, 0.06865034, 0.06836042, 0.07058824, 0.07685969,
        0.08712902, 0.09512474, 0.10942245, 0.14717327, 0.21319905,
        0.31894408, 0.44005493, 0.5972076 , 0.73974212, 0.79563592,
        0.80424201, 0.77680629, 0.72439155, 0.66315709, 0.6205539 ,
        0.59136339, 0.55407034, 0.51183337, 0.48126955, 0.45905241,
        0.44359503, 0.39360647, 0.33247883, 0.29878691, 0.28598459],
       [0.07661555, 0.07098497, 0.07164111, 0.07847715, 0.08960098,
        0.10205234, 0.11189441, 0.13377584, 0.17953765, 0.24435798,
        0.37021439, 0.53138018, 0.71470207, 0.86723125, 0.93109026,
        0.97068742, 0.9206836 , 0.82059968, 0.77393759, 0.72518502,
        0.69723049, 0.63820859, 0.57349508, 0.54044404, 0.5125658 ,
        0.48104067, 0.42185092, 0.3650721 , 0.32027161, 0.30070954],
       [0.0761883 , 0.07347219, 0.07655451, 0.0873579 , 0.104448  ,
        0.12504768, 0.13238727, 0.14227512, 0.17976654, 0.25525292,
        0.39641413, 0.58200961, 0.79380484, 0.94383154, 1.        ,
        1.        , 1.        , 0.94389258, 0.88152895, 0.85290303,
        0.77772183, 0.72477302, 0.64327459, 0.60082399, 0.53676661,
        0.48607614, 0.42684062, 0.37242695, 0.32503243, 0.29857328],
       [0.0804303 , 0.07548638, 0.08142214, 0.09590295, 0.11879149,
        0.14137484, 0.14877546, 0.1485771 , 0.17628748, 0.25600061,
        0.39954223, 0.59349966, 0.82539101, 0.98010224, 1.        ,
        0.9956817 , 1.        , 0.99919127, 0.98258946, 0.9612726 ,
        0.8808423 , 0.8037995 , 0.70806439, 0.64277104, 0.59205005,
        0.52112612, 0.46559854, 0.40114443, 0.34848554, 0.30846113],
       [0.08137636, 0.07981994, 0.08737316, 0.10576028, 0.12225528,
        0.14230564, 0.14819562, 0.142916  , 0.15841917, 0.21976043,
        0.3707332 , 0.57045853, 0.80846876, 0.97276265, 0.99960327,
        0.99920653, 0.99819944, 0.99978637, 0.99832151, 0.96365301,
        0.90923934, 0.81765469, 0.71725032, 0.66317235, 0.60846876,
        0.53975738, 0.466392  , 0.40297551, 0.33667506, 0.29703212],
       [0.07716487, 0.07788205, 0.08696117, 0.1032578 , 0.11734188,
        0.13397421, 0.1363241 , 0.13612573, 0.16073854, 0.23685054,
        0.39136339, 0.58260472, 0.80894179, 0.96975662, 1.        ,
        1.        , 1.        , 1.        , 1.        , 0.98011749,
        0.91194018, 0.85104143, 0.73804837, 0.67684443, 0.60799573,
        0.53466087, 0.45799954, 0.3849546 , 0.33391318, 0.28957046],
       [0.07530327, 0.07371633, 0.08041505, 0.09617762, 0.1160296 ,
        0.13000687, 0.13362325, 0.13542382, 0.17042802, 0.2531014 ,
        0.39249256, 0.55681697, 0.75895323, 0.9013962 , 0.96737621,
        0.9741207 , 0.97988861, 0.97930877, 0.96166934, 0.93058671,
        0.8920119 , 0.82078279, 0.72402533, 0.66617838, 0.60035096,
        0.51851682, 0.4548867 , 0.38654154, 0.33800259, 0.29245441],
       [0.07612726, 0.07347219, 0.07942321, 0.09263752, 0.1111162 ,
        0.12732128, 0.14398413, 0.16021973, 0.20712596, 0.30276951,
        0.44501411, 0.61261921, 0.77622644, 0.86001373, 0.93060197,
        0.95735103, 0.93022049, 0.90074006, 0.89314107, 0.90177768,
        0.8677348 , 0.78344396, 0.6966659 , 0.61287861, 0.56684215,
        0.49781033, 0.43706416, 0.35625238, 0.30019074, 0.26782635],
       [0.07536431, 0.07389944, 0.07913329, 0.09329366, 0.10745403,
        0.13299763, 0.16777295, 0.20903334, 0.28087282, 0.3669337 ,
        0.51096361, 0.65160601, 0.76145571, 0.82662699, 0.84771496,
        0.84402228, 0.81643397, 0.80097658, 0.80257877, 0.82801556,
        0.80445563, 0.7347982 , 0.64538033, 0.58434424, 0.53020523,
        0.47002365, 0.40288396, 0.33575952, 0.28700694, 0.25340658],
       [0.07156481, 0.07115282, 0.07252613, 0.08186465, 0.0985275 ,
        0.13136492, 0.17587549, 0.24162661, 0.33498131, 0.42914473,
        0.56014343, 0.65604639, 0.73432517, 0.77367819, 0.76427863,
        0.71807431, 0.69190509, 0.70617227, 0.72240787, 0.71902037,
        0.70576028, 0.66306554, 0.57520409, 0.50041962, 0.45349813,
        0.40636301, 0.35544366, 0.30244907, 0.26001373, 0.23016709],
       [0.07124437, 0.07095445, 0.07251087, 0.07920958, 0.1025864 ,
        0.14584573, 0.197467  , 0.26108186, 0.3662623 , 0.47635615,
        0.58738079, 0.67333486, 0.72286564, 0.7260853 , 0.69083696,
        0.65348287, 0.62357519, 0.63982605, 0.65975433, 0.67405203,
        0.65381857, 0.60027466, 0.54483864, 0.47833982, 0.44098573,
        0.40209049, 0.33618677, 0.27406729, 0.24048219, 0.22073701],
       [0.07092393, 0.07154955, 0.0743267 , 0.08226139, 0.09831388,
        0.13495079, 0.19009689, 0.26419471, 0.35919738, 0.44888991,
        0.5562829 , 0.62475013, 0.65661097, 0.6290837 , 0.58365759,
        0.55210193, 0.54930953, 0.53910124, 0.55684749, 0.58478676,
        0.57051957, 0.51674678, 0.46550698, 0.42099641, 0.38222324,
        0.34908064, 0.29947356, 0.25461204, 0.2242771 , 0.2061799 ],
       [0.07499809, 0.07765316, 0.08079652, 0.08564889, 0.0952163 ,
        0.11625849, 0.16472114, 0.23291371, 0.32831312, 0.42885481,
        0.50887312, 0.56369879, 0.57940032, 0.54755474, 0.52092775,
        0.48970779, 0.48126955, 0.4908217 , 0.51125353, 0.51749447,
        0.50835431, 0.47421988, 0.42923629, 0.38538186, 0.34389258,
        0.2983444 , 0.26189059, 0.22009613, 0.19946593, 0.18638895],
       [0.07899596, 0.08276493, 0.09018082, 0.09967193, 0.11074998,
        0.12715343, 0.15956359, 0.20015259, 0.28473335, 0.36325628,
        0.42769512, 0.49156939, 0.48667124, 0.49150835, 0.48023194,
        0.45180438, 0.46904707, 0.47782101, 0.5020676 , 0.51091783,
        0.47985046, 0.44885939, 0.40596628, 0.3699855 , 0.33055619,
        0.28464179, 0.24441901, 0.20639353, 0.18974594, 0.1747921 ],
       [0.08130007, 0.0888838 , 0.10243381, 0.11218433, 0.12213321,
        0.12898451, 0.1445182 , 0.18014801, 0.24138247, 0.30704204,
        0.36232547, 0.40610361, 0.41890593, 0.4293431 , 0.42537575,
        0.40428779, 0.40729381, 0.42523842, 0.44701305, 0.45525292,
        0.44818799, 0.41069657, 0.35251392, 0.33475242, 0.30151827,
        0.26228733, 0.21596094, 0.18770123, 0.18266575, 0.15748836],
       [0.08314641, 0.08966201, 0.1038529 , 0.1107805 , 0.12031739,
        0.13528649, 0.14305333, 0.1719234 , 0.21672389, 0.27698177,
        0.31801328, 0.35962463, 0.38422217, 0.40181582, 0.39139391,
        0.38448157, 0.39578851, 0.41803616, 0.44101625, 0.45517662,
        0.45197223, 0.41373312, 0.3738613 , 0.33105974, 0.28685435,
        0.24032959, 0.21120012, 0.19064622, 0.16681163, 0.15185779],
       [0.08151369, 0.08697642, 0.09784085, 0.10983444, 0.11778439,
        0.13374533, 0.15046921, 0.16771191, 0.19443046, 0.24798962,
        0.28975357, 0.34166476, 0.37555505, 0.40199893, 0.41164263,
        0.39714656, 0.41269551, 0.4250248 , 0.44327459, 0.46759747,
        0.44626535, 0.42078279, 0.3816434 , 0.32231632, 0.28938735,
        0.24045167, 0.22255283, 0.1972076 , 0.17541772, 0.15562676],
       [0.0780499 , 0.08300908, 0.09153887, 0.09974823, 0.10818647,
        0.1238117 , 0.14135958, 0.16148623, 0.1837644 , 0.22835126,
        0.27309071, 0.31596857, 0.34168002, 0.36893263, 0.38204013,
        0.38478676, 0.38353552, 0.39942016, 0.41300069, 0.42658122,
        0.42684062, 0.39345388, 0.33817044, 0.28982986, 0.26004425,
        0.22218662, 0.20010681, 0.18861677, 0.171664  , 0.15370413]]

from skimage.feature import peak_local_max
import numpy as np
import matplotlib.pyplot as plt
from skimage import img_as_float

data = img_as_float(img)
coords = peak_local_max(data, min_distance=10,threshold_rel=0.1)

plt.figure(1)
plt.clf()
plt.imshow(data)
plt.plot(coords[:, 1],coords[:, 0],'rx')

rfezzani · 2022-01-04T14:47:33Z

Which version of skimage are you using @jonaszeuner? The behavior you observed was fixed 😉! Here is what I obtain with current version

rfezzani · 2022-01-04T14:49:42Z

BTW, If you don't mind, I will convert this issue to a Discussion.

grlee77 · 2022-04-08T20:53:33Z

comment moved back to issues from Discussion

@jonaszeuner on Jan 4
Great thanks! Indeed I was running 0.16 and updating to 0.19.1 fixed the issue.

taroko-mooncake changed the title ~~peak_local_max attribute min_distance not working~~ peak_local_max attribute min_distance issue Oct 11, 2021

taroko-mooncake changed the title ~~peak_local_max attribute min_distance issue~~ peak_local_max attribute 'min_distance' issue Oct 11, 2021

alexdesiqueira added the type: bug label Oct 11, 2021

rfezzani added the ❓ Needs info Needs more information or investigation label Oct 20, 2021

rfezzani added 🫂 Support User help and QA and removed type: bug labels Oct 22, 2021

scikit-image locked and limited conversation to collaborators Jan 4, 2022

rfezzani converted this issue into a discussion Jan 4, 2022

grlee77 mentioned this issue Feb 20, 2022

2021's calendar of community management #5169

Closed

scikit-image unlocked this conversation Apr 8, 2022

This issue was closed.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

peak_local_max attribute 'min_distance' issue #5598

peak_local_max attribute 'min_distance' issue #5598

taroko-mooncake commented Oct 11, 2021 •

edited

alexdesiqueira commented Oct 11, 2021

taroko-mooncake commented Oct 12, 2021 •

edited

rfezzani commented Oct 12, 2021

taroko-mooncake commented Oct 12, 2021 •

edited

rfezzani commented Oct 22, 2021 •

edited

jonaszeuner commented Dec 30, 2021 •

edited by lagru

rfezzani commented Jan 4, 2022

rfezzani commented Jan 4, 2022

grlee77 commented Apr 8, 2022

peak_local_max attribute 'min_distance' issue #5598

peak_local_max attribute 'min_distance' issue #5598

Comments

taroko-mooncake commented Oct 11, 2021 • edited

Description

Way to reproduce

Version information

alexdesiqueira commented Oct 11, 2021

taroko-mooncake commented Oct 12, 2021 • edited

rfezzani commented Oct 12, 2021

taroko-mooncake commented Oct 12, 2021 • edited

rfezzani commented Oct 22, 2021 • edited

jonaszeuner commented Dec 30, 2021 • edited by lagru

rfezzani commented Jan 4, 2022

rfezzani commented Jan 4, 2022

grlee77 commented Apr 8, 2022

comment moved back to issues from Discussion

taroko-mooncake commented Oct 11, 2021 •

edited

taroko-mooncake commented Oct 12, 2021 •

edited

taroko-mooncake commented Oct 12, 2021 •

edited

rfezzani commented Oct 22, 2021 •

edited

jonaszeuner commented Dec 30, 2021 •

edited by lagru