Research On Marine Infrared Target Detection Technology

In recent years,the demand for marine traffic and operations has increased day by day.At present,marine infrared target recognition and unmanned ship obstacle avoidance technology have become the hotspots of research in various countries.Compared with the visible imaging system,the infrared imaging system has the advantages of excellent environmental adaptability,strong ability to identify camouflage targets,and all-weather use.However,the low signal-to-noise ratio and low contrast make it more complicated in the process of detecting sea targets.For example,the type of background noise is complex,the target occupies few pixels,and the weak target is easily submerged in the noise.The capture accuracy and stability of the detection system will still be seriously affected.In view of the above difficulties,this paper proposes a marine infrared image preprocessing algorithm based on guided filtering for the type of marine infrared image noise,we introduce an adaptive weight based on gradient value and optimizes the algorithm,and then combined with the relevant target tracking algorithm,the target is captured and positioned.We use MATLAB as the platform to simulate the algorithm,and sample the video frame by frame to obtain the image sequence,which verifies the tracking performance of the algorithm.It can be seen from the simulation results that the optimized algorithm can achieve a signal-to-noise ratio increase of 10 dB compared to typical image pre-processing algorithms other than guided filtering.In this paper,FPGA is used as the driving chip of the image acquisition unit,which is called field programmable gate array,and a feasible hardware system is designed through research.In the software design process,based on the consideration of the hardware FPGA,the ISE integrated development environment is fully applied.The software for development and design generally includes four modules: image acquisition,image cache,image algorithm,and image display output.As the core processing chip of this system,at first,FPGA completes the real-time collection of the output image of the infrared camera,and then pre-processes the collected image,which improves the accuracy of the tracking algorithm for the target the image,finally,a tracking algorithm is run based on the processing result image to complete the detection and capture of the target.We built a hardware experiment platform,and experimented with the marine unmanned ship system as the platform.The experimental results show that the combination of the preprocessing algorithm and the Kernelized Correlation Filters tracking algorithm in this paper to achieve target detection and positioning has strong robustness and filtering.After filtering,the tracking position basically coincides with the real position,and the accuracy of detecting the target is higher than 98%.