mirror of
https://github.com/wassname/scikit-image.git
synced 2026-06-29 13:15:47 +08:00
Nelle Varoquaux
117 lines
3.4 KiB
Cython
117 lines
3.4 KiB
Cython
"""
|
|
:author: Damian Eads, 2009
|
|
:license: modified BSD
|
|
"""
|
|
|
|
from __future__ import division
|
|
import numpy as np
|
|
|
|
cimport numpy as np
|
|
cimport cython
|
|
from cpython cimport bool
|
|
|
|
STREL_DTYPE = np.uint8
|
|
ctypedef np.uint8_t STREL_DTYPE_t
|
|
|
|
IMAGE_DTYPE = np.uint8
|
|
ctypedef np.uint8_t IMAGE_DTYPE_t
|
|
|
|
cdef inline int int_max(int a, int b): return a if a >= b else b
|
|
cdef inline int int_min(int a, int b): return a if a <= b else b
|
|
|
|
@cython.boundscheck(False)
|
|
def dilate(np.ndarray[IMAGE_DTYPE_t, ndim=2] image not None,
|
|
np.ndarray[IMAGE_DTYPE_t, ndim=2] selem not None,
|
|
np.ndarray[IMAGE_DTYPE_t, ndim=2] out,
|
|
bool shift_x, bool shift_y):
|
|
cdef int hw = selem.shape[0] // 2
|
|
cdef int hh = selem.shape[1] // 2
|
|
if shift_x:
|
|
hh -= 1
|
|
if shift_y:
|
|
hw -= 1
|
|
|
|
cdef int width = image.shape[0], height = image.shape[1]
|
|
if out is None:
|
|
out = np.zeros([width, height], dtype=IMAGE_DTYPE)
|
|
|
|
assert out.shape[0] == image.shape[0]
|
|
assert out.shape[1] == image.shape[1]
|
|
|
|
cdef int x, y, ix, iy, cx, cy
|
|
cdef IMAGE_DTYPE_t max_so_far
|
|
|
|
cdef int sw = selem.shape[0], sh = selem.shape[1]
|
|
|
|
cdef np.ndarray[np.int_t, ndim=2] xinc = np.zeros([sw, sh], dtype=np.int)
|
|
cdef np.ndarray[np.int_t, ndim=2] yinc = np.zeros([sw, sh], dtype=np.int)
|
|
|
|
for x in range(sw):
|
|
for y in range(sh):
|
|
xinc[x, y] = (x - hw)
|
|
yinc[x, y] = (y - hh)
|
|
|
|
|
|
for x in range(width):
|
|
for y in range(height):
|
|
max_so_far = 0
|
|
for cx in range(0, sw):
|
|
for cy in range(0, sh):
|
|
ix = x + xinc[cx,cy]
|
|
iy = y + yinc[cx,cy]
|
|
if ix>=0 and iy>=0 and ix < width and iy < height \
|
|
and selem[cx, cy] == 1 \
|
|
and image[ix,iy] > max_so_far:
|
|
max_so_far = image[ix,iy]
|
|
out[x,y] = max_so_far
|
|
|
|
return out
|
|
|
|
|
|
@cython.boundscheck(False)
|
|
def erode(np.ndarray[IMAGE_DTYPE_t, ndim=2] image not None,
|
|
np.ndarray[IMAGE_DTYPE_t, ndim=2] selem not None,
|
|
np.ndarray[IMAGE_DTYPE_t, ndim=2] out,
|
|
bool shift_x, bool shift_y):
|
|
cdef int hw = selem.shape[0] // 2
|
|
cdef int hh = selem.shape[1] // 2
|
|
if shift_x:
|
|
hh -= 1
|
|
if shift_y:
|
|
hw -= 1
|
|
|
|
cdef int width = image.shape[0], height = image.shape[1]
|
|
if out is None:
|
|
out = np.zeros([width, height], dtype=IMAGE_DTYPE)
|
|
|
|
assert out.shape[0] == image.shape[0]
|
|
assert out.shape[1] == image.shape[1]
|
|
|
|
cdef int x, y, ix, iy, cx, cy
|
|
cdef IMAGE_DTYPE_t min_so_far
|
|
|
|
cdef int sw = selem.shape[0], sh = selem.shape[1]
|
|
|
|
cdef np.ndarray[np.int_t, ndim=2] xinc = np.zeros([sw, sh], dtype=np.int)
|
|
cdef np.ndarray[np.int_t, ndim=2] yinc = np.zeros([sw, sh], dtype=np.int)
|
|
|
|
for x in range(sw):
|
|
for y in range(sh):
|
|
xinc[x, y] = (x - hw)
|
|
yinc[x, y] = (y - hh)
|
|
|
|
for x in range(width):
|
|
for y in range(height):
|
|
min_so_far = 255
|
|
for cx in range(0, sw):
|
|
for cy in range(0, sh):
|
|
ix = x + xinc[cx,cy]
|
|
iy = y + yinc[cx,cy]
|
|
if ix>=0 and iy>=0 and ix < width \
|
|
and iy < height and selem[cx, cy] == 1 \
|
|
and image[ix,iy] < min_so_far:
|
|
min_so_far = image[ix,iy]
|
|
out[x,y] = min_so_far
|
|
|
|
return out
|