Search results for "vision."

Dynamic Augmented Kalman Filtering for Human Motion Tracking under Occlusion Using Multiple 3D Sensors

2020

In this paper real-time human motion tracking using multiple 3D sensors has been demonstrated in a relatively large industrial robot work cell. The proposed solution extends state-of-the-art by augmenting the constant velocity model and Kalman filter with low-pass filtered velocity states. The presented method is able to handle occlusions by dynamically inclusion in the Kalman filter of only those 3D sensors which provide valid human position data. Human motion tracking was achieved at a frame rate of 20 Hz, with a typical delay of 50 ms to 100 ms and an estimation accuracy of typically 0.10 m to 0.15 m.

GPU-Based Occlusion Minimisation for Optimal Placement of Multiple 3D Cameras

2020

This paper presents a fast GPU-based solution to the 3D occlusion detection problem and the 3D camera placement optimisation problem. Occlusion detection is incorporated into the optimisation problem to return near-optimal positions for 3D cameras in environments containing occluding objects, which maximises the volume that is visible to the cameras. In addition, the authors’ previous work on 3D sensor placement optimisation is extended to include a model for a pyramid-shaped viewing frustum and to take the camera’s pose into account when computing the optimal position.

Visual Marker Guided Point Cloud Registration in a Large Multi-Sensor Industrial Robot Cell

2018

This paper presents a benchmark and accuracy analysis of 3D sensor calibration in a large industrial robot cell. The sensors used were the Kinect v2 which contains both an RGB and an IR camera measuring depth based on the time-of-flight principle. The approach taken was based on a novel procedure combining Aruco visual markers, methods using region of interest and iterative closest point. The calibration of sensors is performed pairwise, exploiting the fact that time-of-flight sensors can have some overlap in the generated point cloud data. For a volume measuring 10m × 14m × 5m a typical accuracy of the generated point cloud data of 5–10cm was achieved using six sensor nodes.

Augmented Reality Visualisation Concepts to Support Intraoperative Distance Estimation

2019

The estimation of distances and spatial relations between surgical instruments and surrounding anatomical structures is a challenging task for clinicians in image-guided surgery. Using augmented reality (AR), navigation aids can be displayed directly at the intervention site to support the assessment of distances and reduce the risk of damage to healthy tissue. To this end, four distance-encoding visualisation concepts were developed using a head-mounted optical see-through AR setup and evaluated by conducting a comparison study. Results suggest the general advantage of the proposed methods compared to a blank visualisation providing no additional information. Using a Distance Sensor concep…

In-line Sorting of Processed Fruit Using Computer Vision

2013

Nowadays, there is a growing demand for quality fruits and vegetables that are simple to prepare and consume, like minimally processed fruits. These products have to accomplish some particular characteristics to make them more attractive to the consumers, like a similar appearance and the total absence of external defects. Although recent advances in machine vision have allowed for the automatic inspection of fresh fruit and vegetables, there are no commercially available equipments for sorting of minority processed fruits, like arils of pomegranate (Punica granatum L) or segments of Satsuma mandarin (Citrus unshiu) ready to eat. This work describes a complete solution based on machine visi…

Mean sets for building 3D probabilistic liver atlas from perfusion MR images

2012

This paper is concerned with liver atlas construction. One of the most important issues in the framework of computational abdominal anatomy is to define an atlas that provides a priori information for common medical task such as registration and segmentation. Unlike other approaches already proposed so far (to our knowledge), in this paper we propose to use the concept of random compact mean set to build probabilistic liver atlases. To accomplish this task a two-tier process was carried out. First a set of 3D images was manually segmented by a physician. We see the different 3D segmented shapes as a realization of a random compact set. Secondly, elements of two known definitions of mean set…

Change-driven Image Architecture on FPGA with adaptive threshold for Optical-Flow Computation

2006

Optical flow computation has been extensively used for object motion estimation in image sequences. However, the results obtained by most optical flow techniques are as accurate as computationally intensive due to the large amount of data involved. A new strategy for image sequence processing has been developed; pixels of the image sequence that significantly change fire the execution of the operations related to the image processing algorithm. The data reduction achieved with this strategy allows a significant optical flow computation speed-up. Furthermore, FPGAs allow the implementation of a custom data-flow architecture specially suited for this strategy. The foundations of the change-dr…

New systems for extracting 3-D shape information from images

1993

Neural architectures may offer an adequate way to deal with early vision since they are able to learn shape features or classify unknown shapes, generalising the features of a few meaningful examples, with a low computational cost after the training phase. Two different neural approaches are proposed by the authors: the first one consists of a cascaded architecture made up by a first stage named BWE (Boundary Webs Extractor) which is aimed to extract a brightness gradient map from the image, followed by a backpropagation network that estimates the geometric parameters of the object parts present in the perceived scene. The second approach is based on the extraction of the boundary webs map …

Corrigendum to “Intelligent agents for feature modelling in computer aided design” [J. Comput. Des. Eng. (2018) 19–40]

2018

Direct 3D Information Determination in an Uncalibrated Stereovision System by Using Evolutionary Algorithms

2011

This paper proposes a 3D panoramic shape reconstruction method based on an uncalibrated stereovision system (USS) composed of five cameras circularly located around the object to be analysed. First, some interesting points are detected from markers placed on the object such that they are visible by two successive cameras of the USS. These points are then matched on both images acquired by a couple of successive cameras. This process is repeated for all the couples of cameras. Second, by using an evolutionary algorithm, the depth values of the different interesting points are calculated. A comparison with a traditional method based on calibrated cameras validates the accuracy of 3D informati…