def find_mask(filename,show=False,outfile="output.jpg", downsample=20, compsizethresh=50,adapthresh=500,blur=10,dilation=10,erosion=1):
p = Pipeline(filename=filename, downsample=20,dilation=10,erosion=1)
p.open() #perform background subtraction
blurnum = int(blur)
blurary = (blurnum,blurnum)
p.blur(blurary) #blur the image
p.crop(factor=8) #crop out the black borders that are produced by the blur
adapthreshnum = int(adapthresh)
p.threshold(adapthreshnum) #do adaptive thresholding
p.erode() #erode the thresholded image
p.connected_components_iterative() #calculate the connected components
"""
try:
p.check_circularity() #check to make sure everything makes sense so far
except:
print "I think this one doesn't have the hole in it..."
#p.save_to_file(p.saved_data,filename=outfile) #saved the masked image to a file
return np.ones(p.data.shape) #return an empty mask
"""
p.select_largest_component() #select the largest connected component
p.connected_components_iterative(full=False) #calculate the connected components of the inverted image
#p.select_largest_component() #selected the largest connected components of the inverted image (this should now be the interior of the hole)
p.select_largest_component()
p.dilate() #dilate the mask
p.mask_original() #mask the original image with the calculated mask
p.save_to_file(out=p.data, filename=outfile) #saved the masked image to a file
if show:
p.show(data=np.concatenate((p.saved_data,p.data),axis=1)) #show the mask for debugging
return p.data
