def _get_roi_cnts(img, x, y, roi_size, thresh, max_area):
worm_img, roi_corner = getWormROI(img, x, y, roi_size)
worm_mask, worm_cnt, _ = getWormMask(worm_img,
thresh,
min_blob_area=max_area
)
if worm_cnt.size > 0:
worm_cnt += roi_corner
return worm_cnt
def _roi2feats(block):
#from a
output = []
for irow, (roi_image, roi_corner) in block:
row_data = trajectories_data.loc[irow]
blob_mask, blob_cnt, _ = getWormMask(
roi_image,
row_data['threshold'],
strel_size,
min_blob_area=row_data['area'] / 2,
is_light_background=is_light_background)
feats = _getBlobFeatures(blob_cnt, blob_mask, roi_image,
roi_corner)
output.append((irow, feats))
return output
def drawThreshMask(self, worm_img, worm_qimg, row_data, read_center=True):
#in very old versions of the tracker I didn't save the area in trajectories table,
#let's assign a default value to deal with this cases
if 'area' in row_data:
min_blob_area = row_data['area'] / 2
else:
min_blob_area = 10
c1, c2 = (row_data['coord_x'],
row_data['coord_y']) if read_center else (-1, -1)
worm_mask, worm_cnt, _ = getWormMask(
worm_img,
row_data['threshold'],
strel_size=self.strel_size,
roi_center_x=c1,
roi_center_y=c2,
min_blob_area=min_blob_area,
is_light_background=self.is_light_background)
worm_mask = QImage(worm_mask.data, worm_mask.shape[1],
worm_mask.shape[0], worm_mask.strides[0],
QImage.Format_Indexed8)
worm_mask = worm_mask.convertToFormat(QImage.Format_RGB32,
Qt.AutoColor)
worm_mask = QPixmap.fromImage(worm_mask)
worm_mask = worm_mask.createMaskFromColor(Qt.black)
p = QPainter(worm_qimg)
p.setPen(QColor(0, 204, 102))
p.drawPixmap(worm_qimg.rect(), worm_mask, worm_mask.rect())
if False:
#test skeletonization
skeleton, ske_len, cnt_side1, cnt_side2, cnt_widths, cnt_area = \
getSkeleton(worm_cnt, np.zeros(0), 49)
for cnt in skeleton, cnt_side1, cnt_side2:
p.setPen(Qt.black)
polyline = QPolygonF()
for point in cnt:
polyline.append(QPointF(*point))
p.drawPolyline(polyline)
p.end()
def getIndCnt(img, x, y, roi_size, thresh, max_area):
worm_img, roi_corner = getWormROI(img, x, y, roi_size)
worm_mask, worm_cnt, _ = getWormMask(worm_img, thresh)
if worm_cnt.size > 0:
worm_cnt += roi_corner
return worm_cnt
#read the data from all the worms in this frame
frame_data = traj_group_by_frame.get_group(frame_number)
#this functions crop all the ROIs of all the worms located in the image
worms_in_frame = getAllImgROI(img, frame_data)
#now i want to pass this into a binary mask with the segmentation
labeled_mask = np.zeros_like(img, dtype=np.uint8)
for iworm, (ind, (worm_img, roi_corner)) in enumerate(worms_in_frame.items()):
row_data = trajectories_data.loc[ind]
#here is where the ROI is really segmented. It uses threshold plus some morphology operations to clean a bit
worm_mask, worm_cnt, cnt_area = getWormMask(worm_img,
row_data['threshold'],
strel_size=5,
min_blob_area=row_data['area'] / 2,
is_light_background = True
)
#now I want to put the ROI mask into the larger labelled image
xi,yi = roi_corner
ss = worm_mask.shape[0]
#here I am saving only the pixels located to the worm.
#This is safer than assigned the whole ROI in case of overlaping regions.
m_roi = labeled_mask[yi:yi+ss, xi:xi+ss].view()
m_roi[worm_mask>0] += iworm
#visualize the results
fig, axs = plt.subplots(1,2, sharex=True, sharey=True)
axs[0].imshow(img)
axs[1].imshow(labeled_mask)
masked_file = "/Volumes/behavgenom_archive$/Avelino/screening/CeNDR/MaskedVideos/CeNDR_Exp_250417/BRC20067_worms10_food1-3_Set4_Pos5_Ch2_25042017_140346.hdf5"
skeletons_file = masked_file.replace('MaskedVideos', 'Results').replace('.hdf5', '_skeletons.hdf5')
with pd.HDFStore(skeletons_file, 'r') as fid:
trajectories_data = fid['/trajectories_data']
roi_generator = generateMoviesROI(masked_file, trajectories_data)
worms_in_frame = next(roi_generator)
for row in worms_in_frame:
worm_img, roi_corner = worms_in_frame[row]
row_data = trajectories_data.loc[row]
worm_mask, worm_cnt, _ = getWormMask(worm_img,
row_data['threshold'],
5)
plt.figure()
plt.subplot(1,2,1)
plt.imshow(worm_img, interpolation=None, cmap='gray')
plt.subplot(1,2,2)
plt.imshow(worm_mask, interpolation=None, cmap='gray')
break
#import h5py
#import pandas as pd
#import numpy as np
#from skimage.filters import threshold_otsu
#from scipy.signal import medfilt