Showing papers on "Orientation (computer vision) published in 2009"

Decomposing a scene into geometric and semantically consistent regions

Abstract: High-level, or holistic, scene understanding involves reasoning about objects, regions, and the 3D relationships between them. This requires a representation above the level of pixels that can be endowed with high-level attributes such as class of object/region, its orientation, and (rough 3D) location within the scene. Towards this goal, we propose a region-based model which combines appearance and scene geometry to automatically decompose a scene into semantically meaningful regions. Our model is defined in terms of a unified energy function over scene appearance and structure. We show how this energy function can be learned from data and present an efficient inference technique that makes use of multiple over-segmentations of the image to propose moves in the energy-space. We show, experimentally, that our method achieves state-of-the-art performance on the tasks of both multi-class image segmentation and geometric reasoning. Finally, by understanding region classes and geometry, we show how our model can be used as the basis for 3D reconstruction of the scene.

Movement recognition as input mechanism

Abstract: The detection of relative motion or orientation between a user and a computing device can be used to control aspects of the device. For example, the computing device can include an imaging element and software for locating positions, shapes, separations, and/or other aspects of a user's facial features relative to the device, such that an orientation of the device relative to the user can be determined. A user then can provide input to the device by performing actions such as tilting the device, moving the user's head, making a facial expression, or otherwise altering an orientation of at least one aspect of the user with respect to the device. Such an approach can be used in addition to, or as an alternative to, conventional input devices such as keypads and touch screens.

Reconstruction of the orientation distribution function in single- and multiple-shell q-ball imaging within constant solid angle

Abstract: q-Ball imaging is a high-angular-resolution diffusion imaging technique that has been proven very successful in resolving multiple intravoxel fiber orientations in MR images. The standard computation of the orientation distribution function (the probability of diffusion in a given direction) from q-ball data uses linear radial projection, neglecting the change in the volume element along each direction. This results in spherical distributions that are different from the true orientation distribution functions. For instance, they are neither normalized nor as sharp as expected and generally require postprocessing, such as artificial sharpening. In this paper, a new technique is proposed that, by considering the solid angle factor, uses the mathematically correct definition of the orientation distribution function and results in a dimensionless and normalized orientation distribution function expression. Our model is flexible enough so that orientation distribution functions can be estimated either from single q-shell datasets or by exploiting the greater information available from multiple q-shell acquisitions. We show that the latter can be achieved by using a more accurate multiexponential model for the diffusion signal. The improved performance of the proposed method is demonstrated on artificial examples and high-angular-resolution diffusion imaging data acquired on a 7-T magnet.

A Shearlet Approach to Edge Analysis and Detection

Abstract: It is well known that the wavelet transform provides a very effective framework for analysis of multiscale edges. In this paper, we propose a novel approach based on the shearlet transform: a multiscale directional transform with a greater ability to localize distributed discontinuities such as edges. Indeed, unlike traditional wavelets, shearlets are theoretically optimal in representing images with edges and, in particular, have the ability to fully capture directional and other geometrical features. Numerical examples demonstrate that the shearlet approach is highly effective at detecting both the location and orientation of edges, and outperforms methods based on wavelets as well as other standard methods. Furthermore, the shearlet approach is useful to design simple and effective algorithms for the detection of corners and junctions.

Object recognition and full pose registration from a single image for robotic manipulation

Abstract: Robust perception is a vital capability for robotic manipulation in unstructured scenes. In this context, full pose estimation of relevant objects in a scene is a critical step towards the introduction of robots into household environments. In this paper, we present an approach for building metric 3D models of objects using local descriptors from several images. Each model is optimized to fit a set of calibrated training images, thus obtaining the best possible alignment between the 3D model and the real object. Given a new test image, we match the local descriptors to our stored models online, using a novel combination of the RANSAC and Mean Shift algorithms to register multiple instances of each object. A robust initialization step allows for arbitrary rotation, translation and scaling of objects in the test images. The resulting system provides markerless 6-DOF pose estimation for complex objects in cluttered scenes. We provide experimental results demonstrating orientation and translation accuracy, as well a physical implementation of the pose output being used by an autonomous robot to perform grasping in highly cluttered scenes.

Orienting displayed elements relative to a user

Abstract: An element is initially displayed on an interactive touch-screen display device with an initial orientation relative to the interactive touch-screen display device. One or more images of a user of the interactive touch-screen display device are captured. The user is determined to be interacting with the element displayed on the interactive touch-screen display device. In addition, an orientation of the user relative to the interactive touch-screen display device is determined based on at least one captured image of the user of the interactive touch-screen display device. Thereafter, in response to determining that the user is interacting with the displayed element, the initial orientation of the displayed element relative to the interactive touch-screen display device is automatically adjusted based on the determined orientation of the user relative to the interactive touch-screen display device.

Facial micro-expressions recognition using high speed camera and 3D-gradient descriptor

Abstract: Facial micro-expressions were proven to be an important behaviour source for hostile intent and danger demeanour detection. In this paper, we present a novel approach for facial micro-expressions recognition in video sequences. First, 200 frame per second (fps) high speed camera is used to capture the face. Second, the face is divided to specific regions, then the motion in each region is recognized based on 3D-Gradients orientation histogram descriptor. For testing this approach, we create a new dataset of facial micro-expressions, that was manually tagged as a ground truth, using a high speed camera. In this work, we present recognition results of 13 different micro-expressions. (6 pages)

What's up CAPTCHA?: a CAPTCHA based on image orientation

Abstract: We present a new CAPTCHA which is based on identifying an image's upright orientation. This task requires analysis of the often complex contents of an image, a task which humans usually perform well and machines generally do not. Given a large repository of images, such as those from a web search result, we use a suite of automated orientation detectors to prune those images that can be automatically set upright easily. We then apply a social feedback mechanism to verify that the remaining images have a human-recognizable upright orientation. The main advantages of our CAPTCHA technique over the traditional text recognition techniques are that it is language-independent, does not require text-entry (e.g. for a mobile device), and employs another domain for CAPTCHA generation beyond character obfuscation. This CAPTCHA lends itself to rapid implementation and has an almost limitless supply of images. We conducted extensive experiments to measure the viability of this technique.

UAV-Borne 3-D Mapping System by Multisensor Integration

Abstract: To represent 3-D space in detail, it is necessary to acquire 3-D shapes and textures simultaneously and efficiently through the use of precise trajectories of sensors. However, there is no reliable, quick, cheap, and handy method for acquiring accurate high-resolution 3-D data on objects in outdoor and moving environments. In this paper, we propose a combination of charge-coupled device cameras, a small and inexpensive laser scanner, an inexpensive inertial measurement unit, and Global Positioning System for a UAV-borne 3-D mapping system. Direct georeferencing is achieved automatically using all of the sensors without any ground control points. A new method of direct georeferencing by the combination of bundle block adjustment and Kalman filtering is proposed. This allows objects to be rendered richly in shape and detailed texture automatically via a UAV from low altitude. This mapping system has been experimentally used in recovery efforts after natural disasters such as landslides, as well as in applications such as river monitoring.

Prospective real-time correction for arbitrary head motion using active markers.

Abstract: Patient motion during an MRI exam can result in major degradation of image quality, and is of increasing concern due to the aging population and its associated diseases. This work presents a general strategy for real-time, intraimage compensation of rigid-body motion that is compatible with multiple imaging sequences. Image quality improvements are established for structural brain MRI acquired during volunteer motion. A headband integrated with three active markers is secured to the forehead. Prospective correction is achieved by interleaving a rapid track-and-update module into the imaging sequence. For every repetition of this module, a short tracking pulse-sequence remeasures the marker positions; during head motion, the rigid-body transformation that realigns the markers to their initial positions is fed back to adaptively update the image-plane—maintaining it at a fixed orientation relative to the head—before the next imaging segment of k-space is acquired. In cases of extreme motion, corrupted lines of k-space are rejected and reacquired with the updated geometry. High-precision tracking measurements (0.01 mm) and corrections are accomplished in a temporal resolution (37 ms) suitable for real-time application. The correction package requires minimal additional hardware and is fully integrated into the standard user interface, promoting transferability to clinical practice. Magn Reson Med, 2009. © 2009 Wiley-Liss, Inc.

Detecting orientation of digital images using face detection information

Abstract: A method of automatically establishing the correct orientation of an image using facial information. This method is based on the exploitation of the inherent property of image recognition algorithms in general and face detection in particular, where the recognition is based on criteria that is highly orientation sensitive. By applying a detection algorithm to images in various orientations, or alternatively by rotating the classifiers, and comparing the number of successful faces that are detected in each orientation, one may conclude as to the most likely correct orientation. Such method can be implemented as an automated method or a semi automatic method to guide users in viewing, capturing or printing of images.

System and method for map matching with sensor detected objects

Abstract: A system and method for map matching with sensor detected objects. A direct sensor and object matching technique can be used to disambiguate objects that the driver passes. The technique also makes it possible for the navigation system to refine (i.e. improve the accuracy of) its position estimate. In some embodiments, a camera in the car can be used to produce, dynamically in real time, images of the vicinity of the vehicle. Map and object information can then be retrieved from a map database, and superimposed on those images for viewing by the driver, including accurately defining the orientation or the platform so that the alignment of the map data and the image data is accurate. Once alignment is achieved, the image can be further enhanced with information retrieved from the database about any in-image objects. Objects may be displayed accurately on a map display as icons that help the driver as he/she navigates the roads.

Performance Analysis of the SIFT Operator for Automatic Feature Extraction and Matching in Photogrammetric Applications

Abstract: In the photogrammetry field, interest in region detectors, which are widely used in Computer Vision, is quickly increasing due to the availability of new techniques. Images acquired by Mobile Mapping Technology, Oblique Photogrammetric Cameras or Unmanned Aerial Vehicles do not observe normal acquisition conditions. Feature extraction and matching techniques, which are traditionally used in photogrammetry, are usually inefficient for these applications as they are unable to provide reliable results under extreme geometrical conditions (convergent taking geometry, strong affine transformations, etc.) and for bad-textured images. A performance analysis of the SIFT technique in aerial and close-range photogrammetric applications is presented in this paper. The goal is to establish the suitability of the SIFT technique for automatic tie point extraction and approximate DSM (Digital Surface Model) generation. First, the performances of the SIFT operator have been compared with those provided by feature extraction and matching techniques used in photogrammetry. All these techniques have been implemented by the authors and validated on aerial and terrestrial images. Moreover, an auto-adaptive version of the SIFT operator has been developed, in order to improve the performances of the SIFT detector in relation to the texture of the images. The Auto-Adaptive SIFT operator (A2 SIFT) has been validated on several aerial images, with particular attention to large scale aerial images acquired using mini-UAV systems.

A Fastening Tool Tracking System Using an IMU and a Position Sensor With Kalman Filters and a Fuzzy Expert System

Abstract: This paper utilizes an intelligent system which incorporates Kalman filters (KFs) and a fuzzy expert system to track the tip of a fastening tool and to identify the fastened bolt. This system employs one inertial measurement unit and one position sensor to determine the orientation and the center of mass location of the tool. KFs are used to estimate the orientation of the tool and the center of mass location of the tool. Although a KF is used for the orientation estimation, orientation error increases over time due to the integration of angular velocity error. Therefore, a methodology to correct the orientation error is required when the system is used for an extended period of time. This paper proposes a method to correct the tilt angle and orientation errors using a fuzzy expert system. When a tool fastens a bolt, the system identifies the fastened bolt using a fuzzy expert system. Through this bolt identification step, the 3-D orientation error of the tool is corrected by using the location and orientation of the fastened bolt and the position sensor outputs. Using the orientation correction method will, in turn, result in improved reliability in determining the tool tip location. The fastening tool tracking system was experimentally tested in a lab environment, and the results indicate that such a system can successfully identify the fastened bolts.

Selection and Orientation of Directional Sensors for Coverage Maximization

Abstract: Sensor nodes may be equipped with a "directional" sensing device (such as a camera) which senses a physical phenomenon in a certain direction depending on the chosen orientation. In this article, we address the problem of selection and orientation of such directional sensors with the objective of maximizing coverage area. Prior works on sensor coverage have largely focused on coverage with sensors that are associated with a unique sensing region. In contrast, directional sensors have multiple sensing regions associated with them, and the orientation of the sensor determines the actual sensing region. Thus, the coverage problems in the context of directional sensors entails selection as well as orientation of sensors needed to activate in order to maximize/ensure coverage. In this article, we address the problem of selecting a minimum number of sensors and assigning orientations such that the given area (or set of target points) is k-covered (i.e., each point is covered k times). The above problem is NP-complete, and even NP-hard to approximate. Thus, we design a simple greedy algorithm that delivers a solution that k-covers at least half of the target points using at most M log(k|C|) sensors, where |C| is the maximum number of target points covered by a sensor and M is the minimum number of sensor required to k-cover all the given points. The above result holds for almost arbitrary sensing regions. We design a distributed implementation of the above algorithm, and study its performance through simulations. In addition to the above problem, we also look at other related coverage problems in the context of directional sensors, and design similar approximation algorithms for them.

Orientation based control of mobile device

Abstract: A mobile device (e.g. a smartphone) controls a feature of a non-navigation related application (e.g. arrangement of menu items, function controlled by a quick launch key, etc.) based on the orientation of the device. These controls may be context-specific such that the orientation is only used to control the feature in particular contexts. Context may be implied from various factors such as time and date information, the location of the device, other devices in proximity to the device, etc. The particular commands to be controlled based on the orientation can be set using the device, or can be loaded from a data file provided from a third party.

Super-resolution orientation estimation and localization of fluorescent dipoles using 3-D steerable filters.

Abstract: Fluorophores that are fixed during image acquisition produce a diffraction pattern that is characteristic of the orientation of the fluorophore’s underlying dipole. Fluorescence localization microscopy techniques such as PALM and STORM achieve super-resolution by applying Gaussian-based fitting algorithms to in-focus images of individual fluorophores; when applied to fixed dipoles, this can lead to a bias in the range of 5–20 nm. We introduce a method for the joint estimation of position and orientation of dipoles, based on the representation of a physically realistic image formation model as a 3-D steerable filter. Our approach relies on a single, defocused acquisition. We establish theoretical, localization-based resolution limits on estimation accuracy using Cramer-Rao bounds, and experimentally show that estimation accuracies of at least 5 nm for position and of at least 2 degrees for orientation can be achieved. Patterns generated by applying the image formation model to estimated position/orientation pairs closely match experimental observations.

Touch screen augmented reality system and method

Abstract: An improved augmented reality (AR) system integrates a human interface and computing system into a single, hand-held device. A touch-screen display and a rear-mounted camera allows a user interact the AR content in a more intuitive way. A database storing graphical images or textual information about objects to be augmented. A processor is operative to analyze the imagery from the camera to locate one or more fiducials associated with a real object, determine the pose of the camera based upon the position or orientation of the fiducials, search the database to find Graphical images or textual information associated with the real object, and display graphical images or textual information in overlying registration with the imagery from the camera.

Systems and Methods for Unifying Coordinate Systems in Augmented Reality Applications

Abstract: Systems and methods for unifying coordinate systems in an augmented reality application or system are disclosed. User devices capture an image of a scene, and determine a location based on the scene image. The scene image may be compared to cartography data or images to determine the location. User devices may propose an origin and orientation or transformation data for a common coordinate system and exchange proposed coordinate system data to agree on a common coordinate system. User devices may also transmit location information to an augmented reality system that then determines an a common coordinate system and transmits coordinate system data such as transformation matrices to the user devices. Images presented to users may be adjusted based on user device locations relative to the coordinate system.

An automatic system of classification of weld defects in radiographic images

Abstract: In this paper, we describe an automatic detection system to recognise welding defects in radiographic images. In a first stage, image processing techniques, including noise reduction, contrast enhancement, thresholding and labelling, were implemented to help in the recognition of weld regions and the detection of weld defects. In a second stage, a set of geometrical features which characterise the defect shape and orientation was proposed and extracted between defect candidates. In a third stage, an artificial neural network (ANN) for weld defect classification was used. With the aim of obtaining the best performance of ANN three different methods for improving network generalisation was used. The results was compared with a method without generalisation. For the input layer, the principal component analysis technique was used in order to reduce the number of feature variables; and, for the hidden layer, a different number of neurons was used.

Three-dimensional nuclear magnetic resonance microimaging of trabecular bone.

Abstract: The conventional approach to measuring structural parameters in trabecular bone rests on stereology from optical images, derived from sections of embedded bone. In order to provide data that are statistically representative of a sufficiently large volume, multiple sections need to be analyzed in each of the three orthogonal planes. In this work, an alternative technique is presented which is based on three-dimensional (3D) volumetric proton nuclear magnetic resonance (NMR) microimaging. The method presented provides images from 9 x 9 x 4 mm 3 volumes of defatted bone specimens in 15-20 minutes scan time at isotropic resolution corresponding to (78 μm) 3 voxel size. Surface-rendered images of bovine and human trabecular bone are shown and an algorithm was developed and implemented for determining the orientation and magnitude of the principal axes of the mean intercept length tensor. (J Bone Miner Res 1995 ;10 :1452-1461)

Astrometry.net: Blind astrometric calibration of arbitrary astronomical images

Abstract: We have built a reliable and robust system that takes as input an astronomical image, and returns as output the pointing, scale, and orientation of that image (the astrometric calibration or WCS information). The system requires no first guess, and works with the information in the image pixels alone; that is, the problem is a generalization of the "lost in space" problem in which nothing--not even the image scale--is known. After robust source detection is performed in the input image, asterisms (sets of four or five stars) are geometrically hashed and compared to pre-indexed hashes to generate hypotheses about the astrometric calibration. A hypothesis is only accepted as true if it passes a Bayesian decision theory test against a background hypothesis. With indices built from the USNO-B Catalog and designed for uniformity of coverage and redundancy, the success rate is 99.9% for contemporary near-ultraviolet and visual imaging survey data, with no false positives. The failure rate is consistent with the incompleteness of the USNO-B Catalog; augmentation with indices built from the 2MASS Catalog brings the completeness to 100% with no false positives. We are using this system to generate consistent and standards-compliant meta-data for digital and digitized imaging from plate repositories, automated observatories, individual scientific investigators, and hobbyists. This is the first step in a program of making it possible to trust calibration meta-data for astronomical data of arbitrary provenance.

Position and orientation measurement apparatus and method thereof

Abstract: A position and orientation measurement apparatus extracts a plurality of geometric feature based on geometric information of an observation object by drawing three-dimensional model data which represents a surface shape of the observation object. Further, the position and orientation measurement apparatus searches an image feature corresponding to the plurality of geometric feature in a reference image in which a position and orientation of an imaging apparatus relative to the observation object has been calculated and selects the geometric feature whose corresponding image feature is detected from the plurality of extracted geometric features. The position and orientation measurement apparatus calculates the position and orientation of the imaging apparatus relative to the observation object by associating the selected geometric feature with an image of the observation object in an input image.

Learning 3-D object orientation from images

Abstract: We propose a learning algorithm for estimating the 3-D orientation of objects. Orientation learning is a difficult problem because the space of orientations is non-Euclidean, and in some cases (such as quaternions) the representation is ambiguous, in that multiple representations exist for the same physical orientation. Learning is further complicated by the fact that most man-made objects exhibit symmetry, so that there are multiple “correct” orientations. In this paper, we propose a new representation for orientations—and a class of learning and inference algorithms using this representation—that allows us to learn orientations for symmetric or asymmetric objects as a function of a single image. We extensively evaluate our algorithm for learning orientations of objects from six categories.

Experiments in place recognition using gist panoramas

Abstract: In this paper we investigate large scale view based localization in urban areas using panoramic images. The presented approach utilizes global gist descriptor computed for portions of panoramic images and a simple similarity measure between two panoramas, which is robust to changes in vehicle orientation, while traversing the same areas in different directions. The global gist feature [14] has been demonstrated previously to be a very effective conventional image descriptor, capturing the basic structure of different types of scenes in a very compact way. We present an extensive experimental validation of our panoramic gist approach on a large scale Street View data set of panoramic images for place recognition or topological localization.

Laser-Based Tracking of Human Position and Orientation Using Parametric Shape Modeling

Abstract: Robots designed to interact socially with people require reliable estimates of human position and motion. Additional pose data such as body orientation may enable a robot to interact more effectively by providing a basis for inferring contextual social information such as people's intentions and relationships. To this end, we have developed a system for simultaneously tracking the position and body orientation of many people, using a network of laser range finders mounted at torso height. An individual particle filter is used to track the position and velocity of each human, and a parametric shape model representing the person's cross-sectional contour is fit to the observed data at each step. We demonstrate the system's tracking accuracy quantitatively in laboratory trials and we present results from a field experiment observing subjects walking through the lobby of a building. The results show that our method can closely track torso and arm movements, even with noisy and incomplete sensor data, and we p...

Image signatures for use in motion-based three-dimensional reconstruction

Abstract: A family of one-dimensional image signatures is obtained to represent each one of a sequence of images in a number of translational and rotational orientations. By calculating these image signatures as images are captured, a new current view can be quickly compared to historical views in a manner that is less dependent on the relative orientation of a target and search image. These and other techniques may be employed in a three-dimensional reconstruction process to generate a list of candidate images from among which full three-dimensional registration may be performed to test for an adequate three-dimensional match. In another aspect this approach may be supplemented with a Fourier-based approach that is selectively applied to a subset of the historical images. By alternating between spatial signatures for one set of historical views and spatial frequency signatures for another set of historical views, a pattern matching system may be implemented that more rapidly reattaches to a three-dimensional model in a variety of practical applications.