Extracting Motion Trajectories Research Articles

A deep learning method for the prediction of 6-DoF ship motions in real conditions

This paper presents a deep learning method for the prediction of ship motions in 6 Degrees of Freedom (DoF). Big data streams of Automatic Identification System (AIS), now-cast, and bathymetry records are used to extract motion trajectories and idealise environmental conditions. A rapid Fluid-Structure Interaction (FSI) model is used to generate ship motions that account for the influence of surrounding water and ship-controlling devices. A transformer neural network that accounts for the influence of operational conditions on ship dynamics is validated by learning the data streams corresponding to ship voyages and hydro-meteorological conditions between two ports in the Gulf of Finland. Predictions for a ship turning circle and motion dynamics between these two ports show that the proposed method can capture the influence of operational conditions on seakeeping and manoeuvring.

영상의 전역 특징과 이동객체의 지역 특징을 융합한 동영상 검색 디스크립터 설계

A descriptor which is suitable for motion analysis by using the motion features of moving objects from the real time image sequence is proposed. To segment moving objects from the background, the background learning is performed. We extract motion trajectories of individual objects by using the sequence of the 1st order moment of moving objects. The center points of each object are managed by linked list. The descriptor includes the 1st order coordinates of moving object belong to neighbor of the pre-defined position in grid pattern, The start frame number which a moving object appeared in the scene and the end frame number which it disappeared. A video retrieval by the proposed descriptor combining global and local feature is more effective than conventional methods which adopt a single feature among global and local features.

Performance studies of four-dimensional cone beam computed tomography

Four-dimensional cone beam computed tomography (4DCBCT) has been proposed to characterize the breathing motion of tumors before radiotherapy treatment. However, when the acquired cone beam projection data are retrospectively gated into several respiratory phases, the available data to reconstruct each phase is under-sampled and thus causes streaking artifacts in the reconstructed images. To solve the under-sampling problem and improve image quality in 4DCBCT, various methods have been developed. This paper presents performance studies of three different 4DCBCT methods based on different reconstruction algorithms. The aims of this paper are to study (1) the relationship between the accuracy of the extracted motion trajectories and the data acquisition time of a 4DCBCT scan and (2) the relationship between the accuracy of the extracted motion trajectories and the number of phase bins used to sort projection data. These aims will be applied to three different 4DCBCT methods: conventional filtered backprojection reconstruction (FBP), FBP with McKinnon–Bates correction (MB) and prior image constrained compressed sensing (PICCS) reconstruction. A hybrid phantom consisting of realistic chest anatomy and a moving elliptical object with known 3D motion trajectories was constructed by superimposing the analytical projection data of the moving object to the simulated projection data from a chest CT volume dataset. CBCT scans with gantry rotation times from 1 to 4 min were simulated, and the generated projection data were sorted into 5, 10 and 20 phase bins before different methods were used to reconstruct 4D images. The motion trajectories of the moving object were extracted using a fast free-form deformable registration algorithm. The root mean square errors (RMSE) of the extracted motion trajectories were evaluated for all simulated cases to quantitatively study the performance. The results demonstrate (1) longer acquisition times result in more accurate motion delineation for each method; (2) ten or more phase bins are necessary in 4DCBCT to ensure sufficient temporal resolution in tumor motion and (3) to achieve the same performance as FBP-4DCBCT with a 4 min data acquisition time, MB-4DCBCT and PICCS-4DCBCT need about 2- and 1 min data acquisition times, respectively.

Extraction of tumor motion trajectories using PICCS-4DCBCT: A validation study

As a counterpart of 4DCT in the treatment planning stage of radiotherapy treatment, 4D cone beam computed tomography (4DCBCT) method has been proposed to verify tumor motion trajectories before radiation therapy treatment delivery. Besides 4DCBCT acquisition using slower gantry rotation speed or multiple rotations, a new method using the prior image constrained compressed sensing (PICCS) image reconstruction method and the standard 1-min data acquisition were proposed. In this paper, the PICCS-4DCBCT method was combined with deformable registration to validate its capability in motion trajectory extraction using physical phantom data, simulated human subject data from 4DCT and in vivo human subject data. Two methods were used to validate PICCS-4DCBCT for the purpose of respiratory motion delineation. The standard 1-min gantry rotation Cone Beam CT acquisition was used for both methods. In the first method, 4DCBCT projection data of a physical motion phantom were acquired using an on-board CBCT acquisition system (Varian Medical Systems, Palo Alto, CA). Using a deformable registration method, the object motion trajectories were extracted from both FBP and PICCS reconstructed 4DCBCT images, and compared against the programmed motion trajectories. In the second method, using a clinical 4DCT dataset, Cone Beam CT projections were simulated by forward projection. Using a deformable registration method, the tumor motion trajectories were extracted from the reconstructed 4DCT and PICCS-4DCBCT images. The performance of PICCS-4DCBCT is assessed against the 4DCT ground truth. The breathing period was varied in the simulation to study its effect on motion extraction. For both validation methods, the root mean square error (RMSE) and the maximum of the errors (MaxE) were used to quantify the accuracy of the extracted motion trajectories. After the validation, a clinical dataset was used to demonstrate the motion delineation capability of PICCS-4DCBCT for human subjects. In both validation studies, the RMSEs of the extracted motion trajectories from PICCS-4DCBCT images are less than 0.7 mm, and their MaxEs are less than 1 mm, for all three directions. In comparison, FBP-4DCBCT shows considerably larger RMSEs in the physical phantom based validation. PICCS-4DCBCT also shows insensitivity to the breathing period in the 4DCT based validation. For the in vivo human subject study, high quality 3D motion trajectory of the tumor was obtained from PICCS-4DCBCT images and showed consistency with visual observation. These results demonstrate accurate delineation of tumor motion trajectory can be achieved using PICCS-4DCBCT and the standard 1-min data acquisition.

TU‐E‐BRC‐07: Accurate Tumor Motion Delineation with 4D Cone Beam CT Using Prior Image Constrained Compressed Sensing (PICCS‐4DCBCT)

Purpose: To validate that 4D cone beam CT using the Prior Image Constrained Compressed Sensing (PICCS) method enables accurate extraction of tumor motion profiles using a 1‐minute data acquisition scheme. Methods: Two methods with known ground truth were used in this validation study. The first method is numerical simulations based on a 4DCT dataset of a human subject, and the second one is a physical motion phantom. In both cases, 600 projections were either simulated from human subject 4DCT data or experimentally acquired using a on‐board CBCT system (Trilogy, Varian Medical Systems, Palo Alto, California) over a 1‐minute gantry rotation through 360°. Images were reconstructed using the PICCS algorithm for all respiratory phases. The motion profile of a selected ROI was extracted using a free‐form deformable registration method. The extracted motion trajectories were compared with the ground truth (the first method uses the motion trajectory from 4DCT images, and the second uses the programmed motion profile) for both methods to evaluate their accuracy. The root mean square error (RMSE) was used to quantify the error of motion trajectory over all phase bins. Results: Results from both methods show excellent agreement between the extracted motion profiles from PICCS‐4DCBCT images and the ground truth. The RMSEs for the first method are 0.68 mm, 0.23 mm and 0.24 mm in x, y and z directions, respectively. The RMSEs for the second method are 0.58 mm and 0.47 mm in AP and SI directions, respectively. The accuracy of motion delineation is at sub‐pixel level. Conclusions: PICCS‐4DCBCT using a standard 1 minute acquisition protocol has been validated to provide tumor motion delineation with high accuracy.

An algorithm to extract three‐dimensional motion by marker tracking in the kV projections from an on‐board imager: four‐dimensional cone‐beam CT and tumor tracking implications

The purpose of this work is to extract three‐dimensional (3D) motion trajectories of internal implanted and external skin‐attached markers from kV cone‐beam projections and reduce image artifact from patient motion in cone‐beam computed tomography (CBCT) from on‐board imager. Cone beam radiographic projections were acquired for a mobile phantom and liver patients with internal implanted and external skin‐attached markers. An algorithm was developed to automatically find the positions of the markers in the projections. It uses normalized cross‐correlation between a template image of a metal seed marker and the projections to find the marker position. From these positions and time‐tagged angular views, the marker 3D motion trajectory was obtained over a time interval of nearly one minute, which is the time required for scanning. This marker trajectory was used to remap the pixels of the projections to eliminate motion. Then, the motion‐corrected projections were used to reconstruct CBCT. An algorithm was developed to extract 3D motion trajectories of internal and external markers from cone‐beam projections using a kV monoscopic on‐board imager. This algorithm was tested and validated using a mobile phantom and patients with liver masses that had radio‐markers implanted in the tumor and attached to the skin. The extracted motion trajectories were used to investigate motion correlation between internal and external markers in liver patients. Image artifacts from respiratory motion were reduced in CBCT reconstructed from cone‐beam projections that were preprocessed to remove motion shifts obtained from marker tracking. With this method, motion‐related image artifacts such as blurring and spatial distortion were reduced, and contrast and position resolutions were improved significantly in CBCT reconstructed from motion‐corrected projections. Furthermore, correlated internal and external marker 3D‐motion tracks obtained from the kV projections might be useful for 4DCBCT, beam gating and tumor motion monitoring or tracking.PACS numbers: 87.57.Q, 87.57.C‐

Extraction and Analysis of Multiple Periodic Motions in Video Sequences

The analysis of periodic or repetitive motions is useful in many applications, such as the recognition and classification of human and animal activities. Existing methods for the analysis of periodic motions first extract motion trajectories using spatial information and then determine if they are periodic. These approaches are mostly based on feature matching or spatial correlation, which are often infeasible, unreliable, or computationally demanding. In this paper, we present a new approach, based on the time-frequency analysis of the video sequence as a whole. Multiple periodic trajectories are extracted and their periods are estimated simultaneously. The objects that are moving in a periodic manner are extracted using the spatial domain information. Experiments with synthetic and real sequences display the capabilities of this approach.

Anatomy of a multicamera video surveillance system

We present a framework for multicamera video surveillance. The framework consists of three phases: detection, representation, and recognition. The detection phase handles multisource spatiotemporal data fusion for efficiently and reliably extracting motion trajectories from video. The representation phase summarizes raw trajectory data to construct hierarchical, invariant, and content-rich descriptions of the motion events. Finally, the recognition phase deals with event classification and identification on the data descriptors. Through empirical study in a parking-lot-surveillance setting, we show that our spatiotemporal fusion scheme and biased sequence-data learning method are highly effective in identifying suspicious events.