Research On Target Tracking Technology Of Unmanned Surface Vehicle Based On Vision

The ocean is rich in resources.In recent years,the country has been committed to the development of the ocean and the protection of the ocean.In marine equipment,unmanned craft has the advantages of fast speed,accurate monitoring and wide application,and is favored by researchers,which has extremely important research significance and practical engineering value.At present,the target tracking of unmanned craft mainly depends on its optical system,but in the face of complex marine environment,such as light change,wave occlusion and so on,it is easy to appear the problems of tracking inaccuracy and target loss.Therefore,in order to meet the needs of more accurate and rapid tracking of targets,this paper based on the traditional correlation filtering tracking technology,improve the sea surface tracking to realize the accurate tracking of sea surface target.In view of the problem that the target tracking is disturbed by external factors such as light change and camera jitter in the sea surface environment,this paper proposes a correlation filtering algorithm for multi-feature fusion,which combines color features on the basis of texture features,and carries out dimensionality reduction processing.The experimental results show that the improved algorithm not only improves the tracking accuracy,reduces the loss probability,but also ensures the performance of the algorithm.In order to solve the problem of tracking failure caused by the change of sea surface target size,this paper first preprocesses the target image,amplifies the image of too small size,and then judges the scale change of the current frame.In order to solve the problem of sea surface target tracking under occlusion,the average peak response difference is used to characterize the similarity between the current target and the filter template.When the parameters are below the threshold,the filter template will not be updated.Kalman algorithm is used to predict the target position.In this paper,the sea surface target tracking video under different scenarios is selected,and the improved algorithm is compared with some existing tracking algorithms.The experimental results show that the accuracy and success rate of this algorithm are among the best.And the tracking speed fully meets the demand of real-time tracking of water unmanned craft.