Research On Perceptual Method Of Near-shore Environment Of Unmanned Boat Based On Machine Vision

With the development of modern intelligent technology,more and more scholars have begun to study the intelligentization of waterway systems.In the future,the inland navigation system will also develop towards unmanned,networked,and technological.In the future development of intelligent shipping,while the efficiency of inland navigation will be greatly improved,the safety of transportation will also be greatly guaranteed.The use of sensors is everywhere in smart shipping.Compared with other sensors,the environment texture obtained by the visual sensor is rich and the amount of information is large.For various environments,the processing algorithm can be flexibly adjusted to achieve the best processing effect.,Can make the unmanned boat more accurate perception of the environment.At present,most researches on the visual perception technology of unmanned boats are focused on the extraction and tracking of water surface targets,and there are few researches on the extraction of waterfront lines and distance perception.This paper focuses on the research of the extraction of the waterfront line and the distance perception of the unmanned boat in the near-shore environment.On the basis of improving the existing method of extraction of the waterfront line,the distance perception of the waterfront line is studied.To measure the distance between water and shore.The main research contents of this article are as follows:(1)Preprocessing of waterfront images.In this paper,by analyzing the waterfront images collected in the field,the characteristics of the waterfront images in the near-shore environment are studied.According to the demand of waterfront sensing,the influence of different color spaces and filtering algorithms on the results of waterfront sensing is analyzed.(2)Waterfront image waterfront extraction.This paper analyzes the current common image segmentation methods.For single-frame waterfront images,a waterfront extraction method based on an improved region growth algorithm is proposed.For the extraction of the waterline of the video sequence,analyze and compare the differences of the commonly used feature point algorithms in the image processing of the waterfront.In the video sequence obtained by the unmanned boat in the near-shore environment,the relationship between the waterline of the adjacent frame in time and space The design uses the method of projection transformation to project the waterline extracted from the previous frame image onto the new frame through projection transformation.Based on this,the waterline extraction can not only greatly reduce the environmental interference,but also reduce the calculation.The amount of water accelerates the extraction of the waterline in the video image sequence.(3)Unmanned boat distance perception in near-shore environment.This paper analyzes the principle of visual perception and starts with the near-shore environment to design a binocular visual ranging scheme to realize the perception of the distance between the water and the shore by the unmanned boat.In this paper,the distance from the shoreline to the unmanned boat is accurately measured by binocular distance measurement,which overcomes the problem of GPS signal loss in the near-shore environment.(4)An experimental environment for unmanned ships was built.The experimental platform is introduced from two aspects of software and hardware.The article concludes with a summary of the full-text work,analyzes the results achieved,and also looks ahead to the future direction of work.