Abstract: Draft survey is a method which is widely used in shipping weighing bulk commodity,At present, artificial observation method is mainly used in draft survey at home and abroad, the ship's water gauge value, liquid tank level of ship, port sea water density are still observed and recorded by manpower. The draft survey has complicated environment, ship tilt, fouling, corrosion and other draft marks objective reasons, and ability of staff observation differences and other subjective reasons, the observation data obtained with great subjectivity, resulting in bilateral trade disputes.According to the draft survey manual observation method can't solve the artificial intervention, artificial errors, low efficiency, many problems such as the dispute, the project is based on the draft survey of automatic measurement as the research core, using sensor technology, wireless communication, image processing of multi subject knowledge,using the method of combining theoretical analysis, optimization algorithm and field test , analysis of different characteristics of the ship's draft video complex environment and the development of draft survey equipment and image recognition software , the project builds a draft survey automatic detection image recognition system. The experimental results show that the system runs stably, with small error and high accuracy. the project thoroughly solves the backward situation of manually observing and recording all the links of draft survey, and finally realizes the scientific, efficient and safe, fair appraisal of the weight of coal carrier in China.