def high_pass(gray, fn, dim=3, save=True, show=True):
# kernal high pass
if dim == 3:
ker = [ [-1.0/16.0, 3.0/16.0,-1.0/16.0],
[-1.0/16.0,16.0/16.0,-1.0/16.0],
[-1.0/16.0,3.0/16.0,-1.0/16.0]]
else:
return None
height, width = gray.shape
ret = np.copy(gray)
for i in range (0, width):
for j in range( 0, height):
if i != 0 and j != 0 and i < width -1 and j < height-1:
val = ker[0][0] * gray[j-1][i-1] + ker[0][1] * gray[j-1][i] + ker[0][2] * gray[j-1][i+1]
val += ker[1][0] * gray[j][i-1] + ker[1][1] * gray[j][i] + ker[1][2] * gray[j][i+1]
val += ker[2][0] * gray[j+1][i-1] + ker[2][1] * gray[j+1][i] + ker[2][2] * gray[j+1][i+1]
if (val < 0):
val = 0
if (val > 1):
val = 1
ret[j][i] = val
if save:
if show:
saveimage(ret, fn, 'high pass', is_gray=True, show=True)
else:
saveimage(ret, fn, 'high pass', is_gray=True, show=False)
return ret
def my_canny(img, fn, sigma=6, with_mask=True, save=True, show=True):
height = img.shape[0]
width = img.shape[1]
if with_mask:
import numpy as np
mask = np.ones((height, width),'uint8')
y1, x1 = 250, 186
y2, x2 = 610, 310
mask[y1: y2, x1: x2] = 0
ret = canny(img, sigma, mask)
else:
ret = canny(img, sigma)
if save:
from src.util import saveimage
if with_mask:
feature = '_sigma' + str(sigma) + '_mask'
else:
feature = '_sigma' + str(sigma)
if show == False:
saveimage(ret, fn+feature, title='canny'+feature, isGray=True, show=False)
else:
saveimage(ret, fn+feature, title='canny'+feature, isGray=True, show=True)
return ret
height = img.size[1]
arr = np.array(img.getdata(), dtype=np.uint8).reshape(height, width, 3)
# get the gray image
from skimage.color import rgb2gray
gray = rgb2gray(arr)
#subtracting the low pass from original
from scipy import ndimage
from src.util import saveimage
npix = 16
npiy = 2
gaussed = ndimage.gaussian_filter(gray, sigma=[npix, npiy])
fn = result_dir + 'gaussian_filter_' + str(npix) + str('_') + str(npiy)
saveimage(gaussed, fn, 'gaussian_filter', is_gray=True, show=False)
high_pass = gray - gaussed
idx = 0 > high_pass
high_pass[idx] = 0
fn = result_dir + 'high_pass_gaussian' + str(npix) + str('_') + str(npiy)
saveimage(high_pass, fn, 'high_pass_gaussian', is_gray=True, show=False)
# binarize the result
idx = high_pass >= 0.05
high_pass[idx] = 1.0
idx = high_pass < 0.05
high_pass[idx] = 0.0
#binary_image = np.array(high_pass, dtype='b')
#high_pass=None
fn = result_dir + 'high_pass_binarize' + str(npix) + str('_') + str(npiy)
height = img.size[1] arr = np.array(img.getdata(), dtype=np.uint8).reshape(height, width, 3) # get the gray image from skimage.color import rgb2gray gray = rgb2gray(arr) arr = None #from src.hist import my_hist #fn = result_dir + 'gray_hist' #my_hist(gray, fn) #from histogram, determine the threshold value min = 0.3, max=0.6 for gray gray_tmp= np.array(gray) gray = None low, high = 0.3, 0.6 idx = low > gray_tmp gray_tmp[idx] = 0 idx = gray_tmp > high gray_tmp[idx] = 0 from src.util import saveimage fn = result_dir + 'thresh_' + str(int(low*10)) + '_' + str(int(high*10)) saveimage(gray_tmp, fn, is_gray=True) from src.morphology import my_opening fn = result_dir + 'threshed_opening' opened = my_opening(gray_tmp, fn, 1, 40) #diff = gray - opened #fn = result_dir + 'threshed gray-opening' #saveimage(diff, fn, is_gray=True)
from src.gabor import gabor_feature print "begin 1" tao = 4 elongation = 8 sigmax = tao/2.35 sigmay = sigmax * elongation max_orientation, max_magnitude, max_frequency = gabor_feature(gray, sigmax=sigmax, sigmay=sigmay) print "end1" from src.gabor import normalize normal_gabor = normalize(max_magnitude) from src.util import saveimage fn = result_dir + 'complex_gabor_magnitude_sigma' + str(int(sigmax)) + '_' + str(int(sigmay)) +'_180' saveimage(normal_gabor, fn, title='gabor magnitude', is_gray=True, show=True) #================================================================== from src.gabor import gabor_feature_real print "begin2" tao = 4 elongation = 8 sigmax = tao/2.35 sigmay = sigmax * elongation max_orientation, max_magnitude, max_frequency = gabor_feature_real(gray, sigmax=sigmax, sigmay=sigmay) print "end2" normal_gabor = normalize(max_magnitude) fn = result_dir + 'real_gabor_magnitude_sigma' + str(int(sigmax)) + '_' + str(int(sigmay)) +'_180' saveimage(normal_gabor, fn, title='gabor magnitude', is_gray=True, show=True)
