def _update_track(xyzf, LW, LH, bounds, meanx, meanz,
vote_grid, carve_grid,
keyvote_grid, keycarve_grid):
# Copy the item
if 0:
occH,vacH = carve.add_votes(xyzf, LW, LH, bounds, meanx, meanz)
vote_grid = np.maximum(np.maximum(occH, vote_grid), keyvote_grid)
carve_grid = np.maximum(np.maximum(vacH, carve_grid), keycarve_grid)
else:
vote_grid = keyvote_grid
carve_grid = keycarve_grid
vote_grid *= (keycarve_grid<=10)
carve_grid *= (keyvote_grid<=10)
result = dict(vote_grid=vote_grid,
carve_grid=carve_grid)
result.update(get_blocks(vote_grid, carve_grid))
return result
def _update_keyframe(xyzf, LW, LH, bounds, meanx, meanz,
keyvote_grid, keycarve_grid,
depthL, depthR, matL, matR, KK,
maskL, maskR, rectL, rectR,
rgbL, rgbR):
def from_rect(m,rect):
(l,t),(r,b) = rect
return m[t:b,l:r]
# Check if we should process a key frame this turn.
global prevstate
if prevstate:
cleancount, rect, depth = prevstate
# Compare the frames
newdepth = from_rect(depthL, rect)
mask = (depth<2047) & (newdepth<2047)
diff = np.abs(depth.astype('f')-newdepth.astype('f'))
score = (diff*mask).mean()
if score < 1.5: cleancount += 1
else: cleancount = 0
# Store the state
prevstate = cleancount, rectL, from_rect(depthL, rectL)
if cleancount == 6:
prevstate = 0, rectL, from_rect(depthL, rectL)
occH,vacH = carve.add_votes(xyzf, LW, LH, bounds, meanx, meanz)
#ccL = colors.project_colors(depthL, rgbL, rectL)
# build the histogram
#bins_
# threshold the histogram
#cH = colors.choose_colors(*ccL)
#ccR = colors.project_colors(depthR, rgbR, rectR)
keycarve_grid = np.maximum(keycarve_grid, vacH)
keyvote_grid = np.maximum(occH,keyvote_grid)
if 0:
HL = carve.carve_background(depthL, LW, LH, bounds, matL, KK)
HR = carve.carve_background(depthR, LW, LH, bounds, matR, KK)
keycarve_grid = np.maximum(keycarve_grid, (HR+HL)*30)
# Merge the grids
keycarve_grid *= (occH<30)
import scipy.ndimage
labels,nlabels = scipy.ndimage.label(keyvote_grid>=30,np.ones((3,3,3)))
for i in range(1,nlabels+1):
if not np.any(labels[:,0,:]==i):
keyvote_grid[labels==i] = 0
result = dict(keyvote_grid=keyvote_grid, keycarve_grid=keycarve_grid)
result.update(_update_track(xyzf, LW, LH, bounds, meanx, meanz,
keyvote_grid, keycarve_grid,
keyvote_grid, keycarve_grid))
return result
else:
return dict(score=score)
else:
prevstate = 0, rectL, from_rect(depthL, rectL)
return {}
