Robust Fast Target Detection And Tracking Algorithms With Scale And Shape Change

As the basic research direction of computer vision,target detection and tracking is also the most popular research field.It is widely used in military scenes,target attack,smart home,video surveillance and traffic management.However,due to the complexity of application scenarios,the limitation of computing platform,the scale and shape changes of targets,the algorithm of target detection and tracking still needs further study.In order to build a real-time target detection and tracking system,this paper studies the above problems,and effectively improves the performance of the first frame detection algorithm and tracking algorithm in tracking system.The main works and innovations of this paper are as follows:(1)First frame detection algorithm based on salient target detection: Saliency detection is an important part of image processing in image processing.It is often used as a key step in the process of image preprocessing,aiming at highlighting visually salient regions or objects in an image.This paper applies the saliency target detection with low computation cost to the first frame detection of the whole tracking system.However,existing saliency detection methods can only generate saliency maps without achieving accuracy,thereby ignoring the integrity of the significant target.The present study proposed a saliency detection algorithm based on the correlation of superpixel and cosine window filtering(SPC)algorithm that introduced a variable weight superpixel segmentation method to keep the edge of information and enhance the integration of the salient target.Space distance and color distance are used to judge the correlation between two superpixels in the proposed algorithm,and based on the superpixel correlation,the saliency map is obtained by all superpixels' voting.When the saliency map is corrected by a cosine window,we call this operation a cosine window filtering.Finally,we used the OSTU method for binarization and obtained the target position and size,using connected domain detection.The experiment results showed that the SPC algorithm can describe superpixels' significance accurately and extract salient object.As compared with other algorithms,our algorithm offers higher accurate results,regarding the accurate position and size of salient targets.(2)Scale change in tracking process: Target scale change has always been an important issue in the field of target tracking.But most existing tracking methods can not calculate the target scale well,resulting in low tracking accuracy.Some scale adaptive algorithms calculate scale by multiple attempts,which greatly improves the computational complexity.For this problem,this paper proposed a new scale adaptive correlation filter tracking algorithm based on the autocorrelation matrix.The method is based on the correlation filter tracker.Firstly,the sample of each frame is constructed as a cyclic matrix,and the kernel recursive least square method is used to learn the classifier.FFT accelerates the convolution process and makes the tracking speed faster.Finally,calculate the autocorrelation matrix using the standard image of each frame during correlation filtering.And get the target scale through the mapping of features between autocorrelation matrix.The experimental results showed that our method can update target scale during real-time tracking and improve the tracking accuracy effectively.Comparing to other algorithms,our algorithm can quickly adapt target scale during tracking and perform better in accuracy and speed.(3)Change of target shape during tracking: In the field of target tracking,the change of target shape is always a difficult problem.Once the scenes video sequence contains shape-deformed target,tracking become a real challenging problem.This paper proposed an algorithm named as Correlation Filtering with Motion Detection(CFMD)to predict lens shaking or camera moving,and to improve the robustness of shape-deformed target tracking.In CMFD,the target position is determined by the weighted outputs of motion detection and correlation filter tracker.We evaluated our CMFD algorithm on the OTB-100 dataset and compared with other target tracking algorithms,including Kernel Correlation Filter(KCF),Scale Adaptive with Multiple Features tracker(SAMF),Discriminative Scale Space Tracker(DSST),and Sum of Template and Pixel-wise LEarners(Staple).The experimental results showed that our algorithm owns the property of robust tracking of shape-deformed targets in video sequences containing lens shaking or camera moving and it achieves the state-of-the-art precision and tracking effects.