Research On Underwater Dynamic Target Recognition And Tracking Technology Based On Binocular Vision

With the continuous development of the world economy,mankind's demand for marine resources continues to increase.For this reason,as the main tool for developing marine resources,the development of underwater robot technology has become more and more important.Underwater vision technology,as the most intuitive way for underwater robots to perceive the external environment,is an important part of underwater robot technology.Therefore,the development of underwater robot vision technology is very necessary.This article mainly studies the binocular vision system,and realizes the recognition and tracking of underwater specific target creatures,and provides more help for underwater robots.The specific research content of this paper is as follows:First of all,starting with the influence of refraction in the underwater environment on camera imaging,after analyzing common underwater binocular imaging systems,an underwater parallel binocular vision model is proposed,and then the reasons for underwater distortion are introduced and analyzed,and finally used This model has completed the calibration of the underwater binocular camera.Secondly,aiming at the problems of uneven illumination,serious dispersion,and more impurity interference in the underwater environment,an improved MSRCP algorithm based on an adaptive algorithm is proposed.The original underwater target image is enhanced by introducing the LMS algorithm to obtain a clear image A,And perform MSR processing on the three color channels of the clear image A,and then combine the three color channels into a processed color image.Finally,the effectiveness of the algorithm is verified through experimental comparison.Thirdly,in view of the difficulty in extracting features of traditional recognition algorithms in the marine environment,an underwater specific target recognition algorithm based on YOLO?v3 is proposed.Train the model through deep learning and identify specific underwater targets,and design experiments to verify the effectiveness of the model.Finally,because traditional tracking methods are difficult to directly apply to the underwater environment,this paper combines the characteristics of the core-related filtering algorithm to solve the problems encountered in the underwater environment,and proposes an underwater target tracking algorithm based on YOLO and nuclear-related filtering,and design experiments to verify the effectiveness of the algorithm.