Depth tracking of occluded ships based on SIFT feature matching

Yadong Liu,Yuesheng Liu,Ziyang Zhong,Yang Chen,Jinfeng Xia,Yunjie Chen 한국인터넷정보학회 2023 KSII Transactions on Internet and Information Syst Vol.17 No.4

Multi-target tracking based on the detector is a very hot and important research topic in target tracking. It mainly includes two closely related processes, namely target detection and target tracking. Where target detection is responsible for detecting the exact position of the target, while target tracking monitors the temporal and spatial changes of the target. With the improvement of the detector, the tracking performance has reached a new level. The problem that always exists in the research of target tracking is the problem that occurs again after the target is occluded during tracking. Based on this question, this paper proposes a DeepSORT model based on SIFT features to improve ship tracking. Unlike previous feature extraction networks, SIFT algorithm does not require the characteristics of pre-training learning objectives and can be used in ship tracking quickly. At the same time, we improve and test the matching method of our model to find a balance between tracking accuracy and tracking speed. Experiments show that the model can get more ideal results.

Robust visual tracking based on global-and-local search with confidence reliability estimation

Fang, Yang,Ko, Seunghyun,Jo, Geun-Sik Elsevier 2019 Neurocomputing Vol.367 No.-

<P><B>Abstract</B></P> <P>Visual object tracking is an open and challenging problem, an online tracker must be able to keep track of the target object for a long time period even in complex scenarios, such as target drift and background occlusion. Discriminative correlation filters (DCF) have shown excellent performance in short-term target tracking problems thanks to their circular dense sampling mechanism and fast computation with a discrete Fourier transform. However, they tend to drift from the target when the target encounters drastic deformation, fast motion, or background occlusion. This can result in a bad model update since the tracker searches the target in a local region centered at the position where target was located in the previous frame. There is no recovery mechanism for target re-identification and re-location. To handle this issue, this paper proposes a global-and-local-search technique that applies a DCF-based tracking model with a novel target-aware detector in a collaborative way. Our tracking model performs the local search process with high tracking confidence, and the target-aware detector is executed to re-identify and locate the target via global search from the entire frame when the model instability and confidence fluctuation are detected by proposed tracking system. Additionally, we designed an enhanced peak-to-sidelobe ratio (EPSR) for confidence estimation, which indicates system instability and fluctuation degree. Thus, the local tracking model and target-aware detector are collaboratively applied for both final target state estimation and online model updates. This not only avoids model corruption from bad updates, but also prevents our tracker from drifting problems for long-term tracking. Experiments on OTB-100 and VOT2016 benchmarks demonstrate that the proposed tracking method achieves state-of-the-art tracking performance in terms of accuracy and robustness, with 22 fps tracking speed (close to realtime) run on a single GPU.</P>

해양환경에서 선박 추적을 위한 라이다를 이용한 궤적 초기화 및 표적 추적 필터

황태현(Tae Hyun Fang),한정욱(Jungwook Han),손남선(Nam-Sun Son),김선영(Sun Young Kim) 제어로봇시스템학회 2016 제어·로봇·시스템학회 논문지 Vol.22 No.2

This paper describes the track initiation and target-tracking filter for ship tracking in a marine environment by using Light Detection And Ranging (LiDAR). LiDAR with three-dimensional scanning capability is more useful for target tracking in the short to medium range compared to RADAR. LiDAR has rotating multi-beams that return point clouds reflected from targets. Through preprocessing the cluster of the point cloud, the center point can be obtained from the cloud. Target tracking is carried out by using the center points of targets. The track of the target is initiated by investigating the normalized distance between the center points and connecting the points. The regular track obtained from the track initiation can be maintained by the target-tracking filter, which is commonly used in radar target tracking. The target-tracking filter is constructed to track a maneuvering target in a cluttered environment. The target-tracking algorithm including track initiation is experimentally evaluated in a sea-trial test with several boats.

추적레이다의 표적 탐지 및 추적 기술 동향

신한섭(Han-Seop Shin),최지환(Jee-Hwan Choi),김대오(Dae-Oh Kim),김태형(Tae-Hyung Kim) 한국항공우주연구원 2009 항공우주산업기술동향 Vol.7 No.1

추적레이다는 안테나로부터 폭이 매우 좁은 펄스를 표적에 위치시켜 표적에서 돌아오는 신호를 수신하여 표적의 위치 (거리, 각도, 속도 등)를 추적하는 장비이다. 추적레이다가 특정한 표적을 탐지하고 추적하기에 앞서 표적과 주변 환경의 특성을 예측하기 위해 잡음 신호와 표적 신호의 수학적 모델이 필요하다. 본 논문에서는 일반적으로 적용되는 잡음 신호와 표적 신호의 모텔에 대한 이론적인 내용을 소개하였고, 이와 더불어 표적의 탐지와 추적을 위한 거리 추적, 각도 추적 및 도플러 주파수 추적에 대한 일반적인 기법들을 기술하였다. In this paper, we described the model of noise, target for tracking radar and range tracking, angle tracking, and Doppler frequency tracking for target acquisition and tracking. Target signal as well as the noise signal is modeled as random process varying with elapsed time. This paper addresses three areas of radar target tracking: range tracking, angle tracking, and Doppler frequency tracking. In general, range tracking is prerequisite to and inherent in both angle and Doppler frequency tracking systems. First, we introduced the several range tracking and described techniques for achieving range tracking. Second, we described the radar angle tracking techniques including conical scan, sequential lobing, and monopulse. Finally, we presented concepts and techniques for Doppler frequency tracking for several radar types.

Visual tracking based Discriminative Correlation Filter Using Target Separation and Detection

Jun-Haeng Lee(이준행) 한국컴퓨터정보학회 2017 韓國컴퓨터情報學會論文誌 Vol.22 No.12

In this paper, we propose a novel tracking method using target separation and detection that are based on discriminative correlation filter (DCF), which is studied a lot recently. ‘Retainability’ is one of the most important factor of tracking. There are some factors making retainability of tracking worse. Especially, fast movement and occlusion of a target frequently occur in image data, and when it happens, it would make target lost. As a result, the tracking cannot be retained. For maintaining a robust tracking, in this paper, separation of a target is used so that normal tracking is maintained even though some part of a target is occluded. The detection algorithm is executed and find new location of the target when the target gets out of tracking range due to occlusion of whole part of a target or fast movement speed of a target. A variety of experiments with various image data sets are conducted. The algorithm proposed in this paper showed better performance than other conventional algorithms when fast movement and occlusion of a target occur.

Dual Detection-Guided Newborn Target Intensity Based on Probability Hypothesis Density for Multiple Target Tracking

( Li Gao ),( Yongjie Ma ) 한국인터넷정보학회 2016 KSII Transactions on Internet and Information Syst Vol.10 No.10

The Probability Hypothesis Density (PHD) filter is a suboptimal approximation and tractable alternative to the multi-target Bayesian filter based on random finite sets. However, the PHD filter fails to track newborn targets when the target birth intensity is unknown prior to tracking. In this paper, a dual detection-guided newborn target intensity PHD algorithm is developed to solve the problem, where two schemes, namely, a newborn target intensity estimation scheme and improved measurement-driven scheme, are proposed. First, the newborn target intensity estimation scheme, consisting of the Dirichlet distribution with the negative exponent parameter and target velocity feature, is used to recursively estimate the target birth intensity. Then, an improved measurement-driven scheme is introduced to reduce the errors of the estimated number of targets and computational load. Simulation results demonstrate that the proposed algorithm can achieve good performance in terms of target states, target number and computational load when the newborn target intensity is not predefined in multi-target tracking systems.

Advances in CRISPR-Cas systems for RNA targeting, tracking and editing

Wang, Fei,Wang, Lianrong,Zou, Xuan,Duan, Suling,Li, Zhiqiang,Deng, Zixin,Luo, Jie,Lee, Sang Yup,Chen, Shi Elsevier 2019 BIOTECHNOLOGY ADVANCES Vol.37 No.5

<P><B>Abstract</B></P> <P>Clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated protein (Cas) systems, especially type II (Cas9) systems, have been widely used in gene/genome targeting. Modifications of Cas9 enable these systems to become platforms for precise DNA manipulations. However, the utilization of CRISPR-Cas systems in RNA targeting remains preliminary. The discovery of type VI CRISPR-Cas systems (Cas13) shed light on RNA-guided RNA targeting. Cas13d, the smallest Cas13 protein, with a length of only ~930 amino acids, is a promising platform for RNA targeting compatible with viral delivery systems. Much effort has also been made to develop Cas9, Cas13a and Cas13b applications for RNA-guided RNA targeting. The discovery of new RNA-targeting CRISPR-Cas systems as well as the development of RNA-targeting platforms with Cas9 and Cas13 will promote RNA-targeting technology substantially. Here, we review new advances in RNA-targeting CRISPR-Cas systems as well as advances in applications of these systems in RNA targeting, tracking and editing. We also compare these Cas protein-based technologies with traditional technologies for RNA targeting, tracking and editing. Finally, we discuss remaining questions and prospects for the future.</P> <P><B>Highlights</B></P> <P> <UL> <LI> RNA targeting and editing are becoming increasingly important </LI> <LI> CRISPR-Cas systems are advancing for RNA targeting, tracking and editing </LI> <LI> The type VI CRISPR-Cas systems are useful for RNA-guided RNA targeting </LI> <LI> Use of Cas9 and Cas13 will advance RNA-targeting technologies </LI> </UL> </P>

클러터밀도 추정 방법 개선을 통한 LM-IPDAF의 표적 추적 성능 향상 연구

유인제(In-Je Yoo),박성제(Sung-Jae Park) 한국산학기술학회 2017 한국산학기술학회논문지 Vol.18 No.5

레이다를 이용한 다수 표적의 상태 추정을 통해 추적 성능을 향상시키는 문제는 중요하다. 클러터 환경에서 추적 필터를 이용하여 다수 표적 추적 시 트랙과 측정치 간의 결합사건이 발생하며 개수가 증가함에 따라 결합사건은 기하급수적으로 증가한다. 이러한 환경에서 다수 표적 추적 필터 설계 시 고려해야할 문제는 첫째, 신속한 거짓트랙 제거 및 표적트랙 확정을 통하여 오경보율 최소화하고, 이를 통해 FTD(False Track Discrimination) 성능을 높인다. 둘째, 다수의 트랙이 측정치를 공유하는 결합사건 발생시 효율적으로 각각의 측정치를 트랙에 할당함으로써 트랙 유지성능을 향상시키는 것이다. 두가지 고려사항을 통해 단일 표적 추적 자료결합 기법을 다수 표적 추적 필터로 확장하여 사용하며, 대표적인 알고리듬으로 JIPDAF(Joint Integrated Probabilistic Data Association Filter)와 LM-IPDAF(Linear Multi-target IPDAF)가 있다. 본 논문에서는 측정치 할당 시 생기는 수 많은 가설들에 대한 확률적 평가를 하지 않음으로써 측정치와 트랙의 개수에 따라 비선형으로 연산량이 증가하지 않으며, 클러터밀도 추정을 통해 트랙을 쇄신하는 트랙존재확률 기반의 LM-IPDAF 알고리듬을 소개한다. 그리고 LM-IPDAF의 트랙존재확률 산출 시 필요한 클러터밀도 추정 방법을 개선함으로써 연산량을 효과적으로 감소시킬 수 있는 방법을 제안하고 시뮬레이션을 통해 기존의 알고리듬과 비교, 분석하여 성능을 검증하였다. 그 결과, 위치 RMSE, Confirmed True Track 측면에서는 동일한 성능을 내면서 시뮬레이션 처리 시간을 약 20% 감소시킬 수 있었다. Improving tracking performance by estimating the status of multiple targets using radar is important. In a clutter environment, a joint event occurs between the track and measurement in multiple target tracking using a tracking filter. As the number increases, the joint event increases exponentially. The problem to be considered when multiple target tracking filter design in such environments is that first, the tracking filter minimizes the rate of false track alarmsby eliminating the false track and quickly confirming the target track. The purpose is to increase the FTD performance. The second consideration is to improve the track maintenance performance by allocating each measurement to a track efficiently when an event occurs. Through two considerations, a single target tracking data association technique is extended to a multiple target tracking filter, and representative algorithms are JIPDAF and LM-IPDAF. In this study, a probabilistic evaluation of many hypotheses in the assignment of measurements was not performed, so that the computation amount does not increase nonlinearly according to the number of measurements and tracks, and the track existence probability based on the track density The LM-IPDAF algorithm was introduced. This paper also proposes a method to reduce the computational complexity by improving the clutter density estimation method for calculating the track existence probability of LM-IPDAF. The performance was verified by a comparison with the existing algorithm through simulation. As a result, it was possible to reduce the simulation processing time by approximately 20% while achieving equivalent performance on the position RMSE and Confirmed True Track.

Multi-target Tracking Algorithm Based on TLD under Dynamic Background

Lixia Xue,Zuocheng Wang,Yanxiang Chen 보안공학연구지원센터 2015 International Journal of Hybrid Information Techno Vol.8 No.7

The tracking of multi-target under dynamic background is a challenging issue in computer vision. We improve TLD algorithm by introducing multi-threading mechanism to expand the number of the tracking targets. Thus the excellent framework of TLD for accurate long-time single target tracking is kept, and the tracking of multiple moving targets is realized at the same time. The time cost of tracking also reduces effectively by this algorithm to meet the real-time demand in many dynamic scenes.

고기동 표적 추적 성능 개선을 위한 연구

정윤식(Yun Sik Jung),김경수(Kyung Su Kim),송택렬(Taek Lyul Song) 제어로봇시스템학회 2010 제어·로봇·시스템학회 논문지 Vol.16 No.11

In this paper, a new realtime algorithm called the RTPBTD-HPDAF (Recursive Temporal Profile Base Target Detection with Highest Probability Data Association Filter) is presented for tracking fast moving small targets with IIR (Imaging Infrared) sensor systems. Spatial filter algorithms are mainly used for target in IIR sensor system detection and tracking however they often generate high density clutter due to various shapes of cloud. The TPBTD (Temporal Profile Base Target Detection) algorithm based on the analysis of temporal behavior of individual pixels is known to have good performance for detection and tracking of fast moving target with suppressing clutter. However it is not suitable to detect stationary and abruptly maneuvering targets. Moreover its computational load may not be negligible. The PTPBTD-HPDAF algorithm proposed in this paper for real-time target detection and tracking is shown to be computationally cheap while it has benefit of tracking targets with abrupt maneuvers. The performance of the proposed RTPBTD-HPDAF algorithm is tested and compared with the spatial filter with HPDAF algorithm for run-time and track initiation at real IIR video.