ESTHER: Joint Camera Self-Calibration and Automatic Radial Distortion Correction From Tracking of Walking Humans

Tang, Zheng; Lin, Yen Shuo; Lee, Kuan Hui; Hwang, Jenq Neng; Chuang, Jen-Hui

doi:10.1109/access.2019.2891224

Cited by 39 publications

(37 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Then, the problem is to estimate a homography given some reference elements (e.g., known objects or manual measurements) extracted from the scene or using the information of detected vanishing points in the image [4], [5], [20], [51], [58], [60], [69], [77], [109], [113]. Another common approach is to calibrate fixed cameras by observing the motion of dynamic objects such as pedestrians [53], [59], [91], [95]. Recently, approaches based on deep learning attempt at estimating directly camera pose and intrinsic parameters on a single image [41], [57].…”

Section: A Scene Geometry Understandingmentioning

confidence: 99%

The Visual Social Distancing Problem

et al. 2020

View full text Add to dashboard Cite

One of the main and most effective measures to contain the recent viral outbreak is the maintenance of the so-called Social Distancing (SD). To comply with this constraint, governments are adopting restrictions over the minimum inter-personal distance between people. Given this actual scenario, it is crucial to massively measure the compliance to such physical constraint in our life, in order to figure out the reasons of the possible breaks of such distance limitations, and understand if this implies a potential threat. To this end, we introduce the Visual Social Distancing (VSD) problem, defined as the automatic estimation of the inter-personal distance from an image, and the characterization of related people aggregations. VSD is pivotal for a non-invasive analysis to whether people comply with the SD restriction, and to provide statistics about the level of safety of specific areas whenever this constraint is violated. We first point out that measuring VSD is not only a geometrical problem, but it also implies a deeper understanding of the social behaviour in the scene. The aim is to truly detect potentially dangerous situations while avoiding false alarms (e.g., a family with children or relatives, an elder with their caregivers), all of this by complying with current privacy policies. We then discuss how VSD relates with previous literature in Social Signal Processing and indicate a path to research new Computer Vision methods that can possibly provide a solution to such problem. We conclude with future challenges related to the effectiveness of VSD systems, ethical implications and future application scenarios.

show abstract

Section: A Scene Geometry Understandingmentioning

confidence: 99%

The Visual Social Distancing Problem

et al. 2020

View full text Add to dashboard Cite

show abstract

“…The formula for specifying the maximum optimization velocity max ( ) v of the particles in this study is shown as (9).…”

Section: ) Proposed Pso With Apammentioning

confidence: 99%

“…Because camera calibration is an important part of image processing, scholars have conducted a lot of research in this field. Camera self-calibration [9], camera plane calibration [10] and Tsai's two-step calibration [11] are relatively mature. In recent years, several other camera calibration methods have emerged [12] [13].…”

Section: Introductionmentioning

confidence: 99%

Diameter Measurement of Fuel Rods Based on Particle Swarm Optimization With All-Parameter Adaptive Mutation Mechanism

Chen

et al. 2020

IEEE Access

View full text Add to dashboard Cite

In nuclear power plants, due to the special working environment, the status of the fuel rods will be changed by the influence of radiation. And these changes, such as diameter expansion, will cause damage to the fuel rods and even affect the safe operation of nuclear power plants. Therefore, it is necessary to accurate diameter measurement for nuclear fuel rods. This paper aims to solve the measurement problem of fuel rods, and proposes a measurement technique based on image processing. In the proposed scheme, based on Zhang's method, tangential distortion is considered to be an affecting factor at first to improve the accuracy of camera calibration. Then, a particle swarm algorithm with all-parameter adaptive mutation method is proposed for the optimization of camera parameters. Finally, a simple and convenient method to calculate pixel equivalent is designed by considering the feature extraction of corner coordinates of subpixel to accurately obtain pixel equivalents. The effectiveness of this measurement scheme is highlighted on a measuring platform. The relative error of the proposed method reaches 0.041%, and RMSE reaches 4.16 μm. Experimental results demonstrate that this approach is efficient for highaccuracy diameter measurement of fuel rods. INDEX TERMS Camera calibration, diameter measurement, image processing, pixel equivalent, parameters optimization, tangential distortion.

show abstract

“…The goal of MOT is to solve the following objective from an input video sequence When the camera parameters are unavailable, we first process a short period of the video sequence by preliminary 2D tracking and foreground segmentation [10]. Each human object is modeled as a pole perpendicular to the ground plane, whose endpoints are located based on the orientation of the foreground blob, from which we can compute the horizon line and vanishing points in the scene for camera self-calibration [28], [29]. An example of the estimated 3D ground plane from camera self-calibration is shown in Fig.…”

Section: A Formulation Of Data Associationmentioning

confidence: 99%

“…Because of the application of camera self-calibration [28], [29], the provided camera matrices in the MOTChallenge 3D benchmark are not adopted in our 3D MOT computation, but are only considered for evaluation. Therefore, our algorithm actually only leverages 2D information for 3D MOT.…”

Section: B Comparison With the 2d State-of-the-artmentioning

confidence: 99%

MOANA: An Online Learned Adaptive Appearance Model for Robust Multiple Object Tracking in 3D

Tang

Hwang

2019

IEEE Access

Self Cite

View full text Add to dashboard Cite

Multiple object tracking has been a challenging field, mainly due to noisy detection sets and identity switch caused by occlusion and similar appearance among nearby targets. Previous works rely on appearance models built on individual or several selected frames for the comparison of features, but they cannot encode long-term appearance changes caused by pose, viewing angle and lighting conditions. In this work, we propose an adaptive model that learns online a relatively long-term appearance change of each target. The proposed model is compatible with any feature of fixed dimension or their combination, whose learning rates are dynamically controlled by adaptive update and spatial weighting schemes. To handle occlusion and nearby objects sharing similar appearance, we also design cross-matching and re-identification schemes based on the application of the proposed adaptive appearance models. Additionally, the 3D geometry information is effectively incorporated in our formulation for data association. The proposed method outperforms all the state-of-the-art on the MOTChallenge 3D benchmark and achieves real-time computation with only a standard desktop CPU. It has also shown superior performance over the state-of-the-art on the 2D benchmark of MOTChallenge.

show abstract

ESTHER: Joint Camera Self-Calibration and Automatic Radial Distortion Correction From Tracking of Walking Humans

Cited by 39 publications

References 34 publications

The Visual Social Distancing Problem

The Visual Social Distancing Problem

Diameter Measurement of Fuel Rods Based on Particle Swarm Optimization With All-Parameter Adaptive Mutation Mechanism

MOANA: An Online Learned Adaptive Appearance Model for Robust Multiple Object Tracking in 3D

Contact Info

Product

Resources

About