First

a binary image containing only land and ocean pixels is generated by intersecting the LiDAR-derived shore digital surface model （DSM） and the ocean surface derived from tidal gauge data. Then a series of image processing algorithms are applied to extract preliminary high and low tide lines. Since LiDAR pulses can not penetrate into water

some regions near the low tide line may have invalid DSM values. In order to overcome this disadvantage

a mobile trend surface fitting method is used to extrapolate the DSM near low water lines

hence more accurate low tide lines can be extracted. Experimental results show that this method can extract high and low tide lines effectively with less manual labors.