Object tracking is a very important topic in the field of computer vision.Many sophisticated appearance models have been proposed.Among them,the trackers based on holistic appearance information provide a compact noti...Object tracking is a very important topic in the field of computer vision.Many sophisticated appearance models have been proposed.Among them,the trackers based on holistic appearance information provide a compact notion of the tracked object and thus are robust to appearance variations under a small amount of noise.However,in practice,the tracked objects are often corrupted by complex noises(e.g.,partial occlusions,illumination variations)so that the original appearance-based trackers become less effective.This paper presents a correntropy-based robust holistic tracking algorithm to deal with various noises.Then,a half-quadratic algorithm is carefully employed to minimize the correntropy-based objective function.Based on the proposed information theoretic algorithm,we design a simple and effective template update scheme for object tracking.Experimental results on publicly available videos demonstrate that the proposed tracker outperforms other popular tracking algorithms.展开更多
The utilization of surgical telementoring has become increasingly prevalent in enhancing the surgical standards of grassroots hospitals in contemporary times.In the traditional framework of surgical telementoring,remo...The utilization of surgical telementoring has become increasingly prevalent in enhancing the surgical standards of grassroots hospitals in contemporary times.In the traditional framework of surgical telementoring,remote doctors guide grassroots doctors via video and audio communication,which may encounter obstacles such as inadequate information transfer and challenges in accurately pinpointing the surgical site and path.To mitigate these issues,this study introduces an intelligent surgical telementoring system based on edge computing.This system enables the transmission of points marked by remote doctors to grassroots doctors,updating these points’coordinates in real-time on the local endoscopic video.In this system,a novel method named P2PTracking(Patchto-Pixel Point Tracking)is implemented.This process begins with the tracking of a square patch surrounding the point marked by the remote doctor using SiameseFC.Following this,feature matching is performed on the tracked patch in the current frame and the square patch in the labelled frame or template frame.The affine transformation is then calculated based on the feature matching results.Lastly,the point tracking result is derived using the computed affine transformation.Experimental results indicate that the proposed system has a transmission speed of 1.99M/s and a transmission latency of 171 ms when transmitting video at a resolution of 1920×1080px,while the proposed method can achieve an accuracy of 96.6%when the pixel error is 4.The code and data are available at https://github.com/hfut66/P2PTracking.git.展开更多
基金This work was supported by National Natural Science Foundation of China(Nos.61702513,61525306,61633021)National Key Research and Development Program of China(No.2016YFB1001000)+1 种基金Capital Science and Technology Leading Talent Training Project(No.Z181100006318030)CAS-AIR and Shandong Provincial Key Research and Development Program(Major Scientific and Technological Innovation Project)(No.2019JZZY010119)。
文摘Object tracking is a very important topic in the field of computer vision.Many sophisticated appearance models have been proposed.Among them,the trackers based on holistic appearance information provide a compact notion of the tracked object and thus are robust to appearance variations under a small amount of noise.However,in practice,the tracked objects are often corrupted by complex noises(e.g.,partial occlusions,illumination variations)so that the original appearance-based trackers become less effective.This paper presents a correntropy-based robust holistic tracking algorithm to deal with various noises.Then,a half-quadratic algorithm is carefully employed to minimize the correntropy-based objective function.Based on the proposed information theoretic algorithm,we design a simple and effective template update scheme for object tracking.Experimental results on publicly available videos demonstrate that the proposed tracker outperforms other popular tracking algorithms.
基金supported by the National Natural Science Foundation of China[grant numbers 62133004,72188101,72293585 and 62322308]the Natural Science Foundation of Anhui Province[grant number 2108085J33]+1 种基金the Anhui Provincial Major Science and Tech-nology Project[grant number 202203a05020010]the Fundamental Research Funds for the Central Universities[grant numbers JZ2023HGQA0125,JZ2024HGTA0174].
文摘The utilization of surgical telementoring has become increasingly prevalent in enhancing the surgical standards of grassroots hospitals in contemporary times.In the traditional framework of surgical telementoring,remote doctors guide grassroots doctors via video and audio communication,which may encounter obstacles such as inadequate information transfer and challenges in accurately pinpointing the surgical site and path.To mitigate these issues,this study introduces an intelligent surgical telementoring system based on edge computing.This system enables the transmission of points marked by remote doctors to grassroots doctors,updating these points’coordinates in real-time on the local endoscopic video.In this system,a novel method named P2PTracking(Patchto-Pixel Point Tracking)is implemented.This process begins with the tracking of a square patch surrounding the point marked by the remote doctor using SiameseFC.Following this,feature matching is performed on the tracked patch in the current frame and the square patch in the labelled frame or template frame.The affine transformation is then calculated based on the feature matching results.Lastly,the point tracking result is derived using the computed affine transformation.Experimental results indicate that the proposed system has a transmission speed of 1.99M/s and a transmission latency of 171 ms when transmitting video at a resolution of 1920×1080px,while the proposed method can achieve an accuracy of 96.6%when the pixel error is 4.The code and data are available at https://github.com/hfut66/P2PTracking.git.
