Research On Key Technologies Of Air Target Recognition Based On Image Processing

With the rapid development of science and technology,digital image processing technology has been widely used in various sectors of the national economy.,Most of the information in life is obtained by means of the eye.Computer vision is equivalent to putting the computer into the human eye.In recent years,the development speed of computer vision has been accelerating,and its application has been widely used in industrial detection,biomedicine,military,aerospace and other fields.Today,there are hardly any technical areas unrelated to vision.However,there are still great challenges.The target recognition technology in computer vision still has many difficulties that need to be solved.This article is based on the stealth electric surface water unmanned boat project based on the study of the use of image acquisition equipment for the acquisition of the target on the surface of the sea,and its corresponding treatment.Part of the stealth electric surface unmanned watercraft project is to develop a visual perception system for unmanned boats,to detect the presence of airborne objects such as airplanes on its driving path,and to provide reference for unmanned trawl trajectory planning to ensure that no one The boat is driving safely.According to the characteristics of images captured by mobile carriers,corresponding image preprocessing was performed.Then,on the basis of this,several commonly used edge detection operators are used for experiment comparison,and the Canny edge detection operator is finally selected.First of all,the pre-processing of acquired sea-sky image mainly includes image smoothing,noise reduction filtering and binarization.Four common edge detection operators such as Robert,Sobel,Prewitt and Gauss-Laplace operators were used to compare the experimental results.Secondly,the application of improved Hough transform find sea antenna,determine the target area by the sea after the antenna rotating accurate extraction of the target.Then,the target image contour extraction to obtain a profile curve airplane image.Extraction of the target geometric features and characteristics of moments,and extracted features improved application methods PCA reduction feature fusion,thecharacteristics of the reduction of input support vector machine learning classifier,and finally complete the identification.Finally,using the software program implemented Halcon write source code,C++ code embedded derived using MFC frame system made by Halcon software package to generate the code DLL dynamic library file,the function module is implemented in the form of calls.Finally,through the verification of system software,the features extracted in the paper are effective and the system operation is simple.Through a large number of experiments,the recognition rate is about 80%.