Abstract: Aimed at the mining environment image of a seabed nodule-collecting vehicle,the detail of the image was enhanced by subsection linear transformation,and the interferences of suspensions were removed with a median filter.The profile of terrains and obstacles was extracted by an anti-noise gradient operator in morphology,and the rate of change of surface brightness was computed by subsection-linear fitting.According to the feature of the brightness variation,the type of obstacles was estimated,and the profile was detailed and linked by self-adapting morphology.Based on the image information of obstacles,the distance,height and width of the obstacles were computed by projection transformation.Close analysis of land images demonstrated the reliability of computing such parameters as position,height and gradient.This method not only reserves the profile information,but also improves the anti-noisy ability and the anti-interconnection ability,detects the deep-seabed terrains and obstacles efficiently,and works out the position and figure efficiently,so it can be used to provide reliable data for the information fusion technology of the obstacle-avoiding system in a nodule-collecting vehicle.