Design And Implementation Of Side Scan Sonar Imaging Simulator Based On Ray Acoustics

Target detection based on sonar images is an important part of the underwater perception,which is widely used in underwater acoustics research.However,the underwater environment is complex,the cost of sonar imaging experiments is high,and the quantity and quality of acoustic images are difficult to be guaranteed.The openness of sonar datasets used for research internationally is limited,and the performance of algorithms based on supervised learning largely depends on the quantity and quality of the collected data.These problems limit the completion of underwater target detection tasks,therefore,further expansion sonar image data integration is one of the important problems to be solved urgently in this field.The existing studies are mostly based on operations such as rotation and translation,which conflict with the principle of sonar imaging,and fail to fully consider the diversity of underwater targets and the physical factors that affect sonar imaging.In view of the above problems,this dissertation designs a simulator of reproducible side-scan sonar equipment by studying the simulation imaging technology based on ray acoustics to obtain sonar images that are close to the physical reality and reduce the difficulty of acquiring imaging data of acoustic targets.Underwater target detection provides data support and has certain practical application value.The main contents of this dissertation are as follows:Firstly,by analyzing the target types of acoustic images,this dissertation proposes the principle of selecting targets objects and using the robot simulation software Gazebo and the three-dimensional engine Open Scene Graph to model the targets and the marine environment,respectively,to construct a virtual underwater scene.Secondly,a real-time side-scan sonar imaging simulator is designed,which combines rasterization and selective ray tracing technology on the graphics processor,and only calculates the surface of the object reflected by the sound to reduce the amount of calculation,and characterizes the reflection as two sonar rendering parameters namely pulse distance and echo intensity.In the algorithm design,the actual physical process is combined with the sound propagation,and the sound attenuation,speckle noise,reverberation and material properties existing in the sonar image are modeled and processed,the parameters are processed into simulated sonar data based on graphics technology in the absence of real data,the simulator can generate simulated side-scan sonar images with diversity,reducing the risk of field experiments.Finally,the simulator is used to generate side-scan sonar images of various underwater objects.This dissertation provides a dataset named Underwater Object Simulation Image.By comparing the simulated images with the sonar images collected by the Songshan Lake experiment,proved that the simulator can obtain images that are close to the physical reality.The phenomenon proves the effectiveness of the simulator;by calculating the time of rendering the image,proved that the simulator can process acoustic data online,thus providing acoustic data support for the "Underwater Agile Robot Collaborative Operation Platform" project.