Study Of Image Recognition Technology On Vision Target Of Unmanned Surface Vessel

In the 21 st century,As the rapidly developing of the computer technology,communication technology and the network technology,The demand for intelligent monitoring equipment is becoming more and more urgent.So,video image processing technology is an unprecedented development by leaps and bounds.USV is short for unmanned surface vessel.It is a motion platform,which can navigate autonomously in the marine environment and complete various tasks.USV can finish corresponding action according to preset function,in the light of the video image taken by digital video of USV,the technology and methods were studied for the surface targets recognition based on this background.There are strict accuracy and rapidity requirements in image processing process for target image recognition technology systems.According to the image feature of common surface targets in the USV visual system,a study for the USV visual target image recognition technology is studied,At the same time a simulation tests their effectiveness analysis.Regarding the common surface targets of the USV as the research object,Find a way with image pre-processing,target segmentation,the analysis and comparison of target characteristics and choice of target identification method.respectively,The image preprocessing and segmentation algorithm,feature selection and extraction method,and the recognition method of targets image are in the study.First of all,considering the unmanned craft under complex marine environment,Pretreatment was done for the collected video images by using the computer,which includes image filtering and image histogram enhancement.And then the target image was obtained by segmenting the image pretreated.Secondly,Different texture characteristics of target were extracted after segmenting the target image,which are used to classifying the ships and non-ships.classify the ships type by taking the secondary combination features with Hu invariant moment,affine invariant moment and geometric features.Finally,The study of the classification and recognition of surface ship target is the key problem.In the field of ship target recognition,according to the geometry features and shape invariance features of ships,The image recognition target method based on LVQ neural network was mainly studied in this paper.At the same time,combining with the MATLAB environment and the recognition methods,the simulation result is promising.