边缘检测算法

#不知道当时写的时候在干嘛，不过效果还可以
def edge(IDX):
	pyris.show_npy(IDX)
	# edge=mm.dilation(mask,selem=[[1,1,1],[1,0,1],[1,1,1]])-mask
	selems=[[[1,0,0],[0,0,0],[0,0,0]],
			[[0,1,0],[0,0,0],[0,0,0]],
			[[0,0,1],[0,0,0],[0,0,0]],
			[[0,0,0],[1,0,0],[0,0,0]],
			[[0,0,0],[0,0,1],[0,0,0]],
			[[0,0,0],[0,0,0],[1,0,0]],
			[[0,0,0],[0,0,0],[0,1,0]],
			[[0,0,0],[0,0,0],[0,0,1]]]
	s2=np.zeros(IDX.shape)
	for selem in selems:
		selem = np.array(selem)
		s2 += (cv2.filter2D(IDX,-1,selem)-IDX)**2
	s2=(s2**0.5)
	print(s2.shape)
	s2=s2/np.abs(IDX)
	print(s2.shape)
	s2=np.where(s2>10,10,s2)
	#pyris.show_npy(s2)
	thresh=threshold_otsu( s2[np.isfinite(s2)] )
	s2=s2>thresh
	pyris.show_npy(s2)
	return s2
edge(pyris.load(r'E:\Projects\BraidedRivers\SRYzR\Reaches\TTR_M\process_TTR_M\IDX\1991_227.npy'))