Showing papers on "Orientation (computer vision) published in 2015"

Sar-sift: a sift-like algorithm for sar images

Abstract: The scale-invariant feature transform (SIFT) algorithm and its many variants are widely used in computer vision and in remote sensing to match features between images or to localize and recognize objects. However, mostly because of speckle noise, it does not perform well on synthetic aperture radar (SAR) images. In this paper, we introduce a SIFT-like algorithm specifically dedicated to SAR imaging, which is named SAR-SIFT. The algorithm includes both the detection of keypoints and the computation of local descriptors. A new gradient definition, yielding an orientation and a magnitude that are robust to speckle noise, is first introduced. It is then used to adapt several steps of the SIFT algorithm to SAR images. We study the improvement brought by this new algorithm, as compared with existing approaches. We present an application of SAR-SIFT to the registration of SAR images in different configurations, particularly with different incidence angles.

Voting for Voting in Online Point Cloud Object Detection

Abstract: This paper proposes an efficient and effective scheme to applying the sliding window approach popular in computer vision to 3D data. Specifically, the sparse nature of the problem is exploited via a voting scheme to enable a search through all putative object locations at any orientation. We prove that this voting scheme is mathematically equivalent to a convolution on a sparse feature grid and thus enables the processing, in full 3D, of any point cloud irrespective of the number of vantage points required to construct it. As such it is versatile enough to operate on data from popular 3D laser scanners such as a Velodyne as well as on 3D data obtained from increasingly popular push-broom configurations. Our approach is “embarrassingly parallelisable” and capable of processing a point cloud containing over 100K points at eight orientations in less than 0.5s. For the object classes car, pedestrian and bicyclist the resulting detector achieves best-in-class detection and timing performance relative to prior art on the KITTI dataset as well as compared to another existing 3D object detection approach.

Occlusion-Aware Depth Estimation Using Light-Field Cameras

Abstract: Consumer-level and high-end light-field cameras are now widely available. Recent work has demonstrated practical methods for passive depth estimation from light-field images. However, most previous approaches do not explicitly model occlusions, and therefore cannot capture sharp transitions around object boundaries. A common assumption is that a pixel exhibits photo-consistency when focused to its correct depth, i.e., all viewpoints converge to a single (Lambertian) point in the scene. This assumption does not hold in the presence of occlusions, making most current approaches unreliable precisely where accurate depth information is most important - at depth discontinuities. In this paper, we develop a depth estimation algorithm that treats occlusion explicitly, the method also enables identification of occlusion edges, which may be useful in other applications. We show that, although pixels at occlusions do not preserve photo-consistency in general, they are still consistent in approximately half the viewpoints. Moreover, the line separating the two view regions (correct depth vs. occluder) has the same orientation as the occlusion edge has in the spatial domain. By treating these two regions separately, depth estimation can be improved. Occlusion predictions can also be computed and used for regularization. Experimental results show that our method outperforms current state-of-the-art light-field depth estimation algorithms, especially near occlusion boundaries.

Orientation robust object detection in aerial images using deep convolutional neural network

Abstract: Detecting objects in aerial images is challenged by variance of object colors, aspect ratios, cluttered backgrounds, and in particular, undetermined orientations. In this paper, we propose to use Deep Convolutional Neural Network (DCNN) features from combined layers to perform orientation robust aerial object detection. We explore the inherent characteristics of DC-NN as well as relate the extracted features to the principle of disentangling feature learning. An image segmentation based approach is used to localize ROIs of various aspect ratios, and ROIs are further classified into positives or negatives using an SVM classifier trained on DCNN features. With experiments on two datasets collected from Google Earth, we demonstrate that the proposed aerial object detection approach is simple but effective.

StructSLAM: Visual SLAM With Building Structure Lines

Abstract: We propose a novel 6-degree-of-freedom (DoF) visual simultaneous localization and mapping (SLAM) method based on the structural regularity of man-made building environments. The idea is that we use the building structure lines as features for localization and mapping. Unlike other line features, the building structure lines encode the global orientation information that constrains the heading of the camera over time, eliminating the accumulated orientation errors and reducing the position drift in consequence. We extend the standard extended Kalman filter visual SLAM method to adopt the building structure lines with a novel parameterization method that represents the structure lines in dominant directions. Experiments have been conducted in both synthetic and real-world scenes. The results show that our method performs remarkably better than the existing methods in terms of position error and orientation error. In the test of indoor scenes of the public RAWSEEDS data sets, with the aid of a wheel odometer, our method produces bounded position errors about 0.79 m along a 967-m path although no loop-closing algorithm is applied.

Quantification of the internal structure and automatic generation of voxel models of textile composites from X-ray computed tomography data

Abstract: X-ray computed tomography provides an opportunity for a detailed examination of the inner structure of fibre reinforced composites. Three-dimensional images, obtained with micro-CT, can be used for a realistic modelling of composite materials. All modelling objectives imply the knowledge of the orientations of the fibres inside the composite, which determine the local (anisotropic) properties. This paper investigates application of the structure tensor, a concept from the image processing field, to the determination of the orientations of fibres and to segment the image into the material’s components, for the purpose of an automatic generation of a voxel-based description of the representative volume element. The segmentation of the images of CFRP materials into its components is performed by thresholding, or by clustering in a two-dimensional parameter space of the degree of anisotropy and average grey value or one of the components of the orientation vector. Clustering allows not only separating the matrix from the yarns, but also distinguishing the yarns of different primary orientations.

Where are they looking

Abstract: Humans have the remarkable ability to follow the gaze of other people to identify what they are looking at. Following eye gaze, or gaze-following, is an important ability that allows us to understand what other people are thinking, the actions they are performing, and even predict what they might do next. Despite the importance of this topic, this problem has only been studied in limited scenarios within the computer vision community. In this paper, we propose a deep neural network-based approach for gaze-following and a new benchmark dataset, GazeFollow, for thorough evaluation. Given an image and the location of a head, our approach follows the gaze of the person and identifies the object being looked at. Our deep network is able to discover how to extract head pose and gaze orientation, and to select objects in the scene that are in the predicted line of sight and likely to be looked at (such as televisions, balls and food). The quantitative evaluation shows that our approach produces reliable results, even when viewing only the back of the head. While our method outperforms several baseline approaches, we are still far from reaching human performance on this task. Overall, we believe that gaze-following is a challenging and important problem that deserves more attention from the community.

Identification of damage in buildings based on gaps in 3D point clouds from very high resolution oblique airborne images

Abstract: Point clouds generated from airborne oblique images have become a suitable source for detailed building damage assessment after a disaster event, since they provide the essential geometric and radiometric features of both roof and facades of the building. However, they often contain gaps that result either from physical damage or from a range of image artefacts or data acquisition conditions. A clear understanding of those reasons, and accurate classification of gap-type, are critical for 3D geometry-based damage assessment. In this study, a methodology was developed to delineate buildings from a point cloud and classify the present gaps. The building delineation process was carried out by identifying and merging the roof segments of single buildings from the pre-segmented 3D point cloud. This approach detected 96% of the buildings from a point cloud generated using airborne oblique images. The gap detection and classification methods were tested using two other data sets obtained with Unmanned Aerial Vehicle (UAV) images with a ground resolution of around 1–2 cm. The methods detected all significant gaps and correctly identified the gaps due to damage. The gaps due to damage were identified based on the surrounding damage pattern, applying Gabor wavelets and a histogram of gradient orientation features. Two learning algorithms – SVM and Random Forests were tested for mapping the damaged regions based on radiometric descriptors. The learning model based on Gabor features with Random Forests performed best, identifying 95% of the damaged regions. The generalization performance of the supervised model, however, was less successful: quality measures decreased by around 15–30%.

Fingerprint Image Enhancement Algorithm and Performance Evaluation

Abstract: Fingerprint identification is one of the most important biometric technologies which has drawn a substantial amount of attention recently. The uniqueness of a fingerprint is exclusively determined by the local ridge characteristics and their relationships. Very important. Fingerprint images get degraded and corrupted due to variations in skin and impression conditions. Thus, image enhancement techniques are employed prior to minutiae extraction. However, the performance of a minutiae extraction algorithm relies heavily on the quality of the input fingerprint images. Here introducing a fast fingerprint enhancement algorithm, which can adaptively improve the clarity of ridge and valley structures of fingerprint images based on the estimated local ridge orientation and frequency and evaluated the performance of the image enhancement algorithm using the goodness index of the extracted minutiae and the accuracy of an online fingerprint verification system.

Reusing 60GHz Radios for Mobile Radar Imaging

Abstract: The future of mobile computing involves autonomous drones, robots and vehicles. To accurately sense their surroundings in a variety of scenarios, these mobile computers require a robust environmental mapping system. One attractive approach is to reuse millimeterwave communication hardware in these devices, e.g. 60GHz networking chipset, and capture signals reflected by the target surface. The devices can also move while collecting reflection signals, creating a large synthetic aperture radar (SAR) for high-precision RF imaging. Our experimental measurements, however, show that this approach provides poor precision in practice, as imaging results are highly sensitive to device positioning errors that translate into phase errors. We address this challenge by proposing a new 60GHz imaging algorithm, {\em RSS Series Analysis}, which images an object using only RSS measurements recorded along the device's trajectory. In addition to object location, our algorithm can discover a rich set of object surface properties at high precision, including object surface orientation, curvature, boundaries, and surface material. We tested our system on a variety of common household objects (between 5cm--30cm in width). Results show that it achieves high accuracy (cm level) in a variety of dimensions, and is highly robust against noises in device position and trajectory tracking. We believe that this is the first practical mobile imaging system (re)using 60GHz networking devices, and provides a basic primitive towards the construction of detailed environmental mapping systems.

Automatic registration of UAV-borne sequent images and LiDAR data

Abstract: Use of direct geo-referencing data leads to registration failure between sequent images and LiDAR data captured by mini-UAV platforms because of low-cost sensors. This paper therefore proposes a novel automatic registration method for sequent images and LiDAR data captured by mini-UAVs. First, the proposed method extracts building outlines from LiDAR data and images and estimates the exterior orientation parameters (EoPs) of the images with building objects in the LiDAR data coordinate framework based on corresponding corner points derived indirectly by using linear features. Second, the EoPs of the sequent images in the image coordinate framework are recovered using a structure from motion (SfM) technique, and the transformation matrices between the LiDAR coordinate and image coordinate frameworks are calculated using corresponding EoPs, resulting in a coarse registration between the images and the LiDAR data. Finally, 3D points are generated from sequent images by multi-view stereo (MVS) algorithms. Then the EoPs of the sequent images are further refined by registering the LiDAR data and the 3D points using an iterative closest-point (ICP) algorithm with the initial results from coarse registration, resulting in a fine registration between sequent images and LiDAR data. Experiments were performed to check the validity and effectiveness of the proposed method. The results show that the proposed method achieves high-precision robust co-registration of sequent images and LiDAR data captured by mini-UAVs.

Object detection and analysis via unmanned aerial vehicle

Abstract: An unmanned aerial vehicle (UAV) can include one or more cameras for capturing image data within a field of view that depends in part upon the location and orientation of the UAV. At least a portion of the image data can be processed on the UAV to locate objects of interest, such as people or cars, and use that information to determine where to fly the drone in order to capture higher quality image data of those or other such objects. Once identified, the objects of interest can be counted, and the density, movement, location, and behavior of those objects identified. This can help to determine occurrences such as traffic congestion or unusual patterns of pedestrian movement, as well as to locate persons, fires, or other such objects. The data can also be analyzed by a remote system or service that has additional resources to provide more accurate results.

Latent orientation field estimation via convolutional neural network

Abstract: The orientation field of a fingerprint is crucial for feature extraction and matching However, estimation of orientation fields in latents is very challenging because latents are usually of poor quality Inspired by the superiority of convolutional neural networks (ConvNets) for various classification and recognition tasks, we pose latent orientation field estimation in a latent patch to a classification problem, and propose a ConvNet based approach for latent orientation field estimation The underlying idea is to identify the orientation field of a latent patch as one of a set of representative orientation patterns To achieve this, 128 representative orientation patterns are learnt from a large number of orientation fields For each orientation pattern, 10,000 fingerprint patches are selected to train the ConvNet To simulate the quality of latents, texture noise is added to the training patches Given image patches extracted from a latent, their orientation patterns are predicted by the trained ConvNet and quilted together to estimate the orientation field of the whole latent Experimental results on NIST SD27 latent database demonstrate that the proposed algorithm outperforms the state-of-the-art orientation field estimation algorithms and can boost the identification performance of a state-of-the-art latent matcher by score fusion

CannyLines: A parameter-free line segment detector

Abstract: In this paper, we present a robust line segment detection algorithm to efficiently detect the line segments from an input image. Firstly a parameter-free Canny operator, named as CannyPF, is proposed to robustly extract the edge map from an input image by adaptively setting the low and high thresholds for the traditional Canny operator. Secondly, both efficient edge linking and splitting techniques are proposed to collect collinear point clusters directly from the edge map, which are used to fit the initial line segments based on the least-square fitting method. Thirdly, longer and more complete line segments are produced via efficient extending and merging. Finally, all the detected line segments are validated due to the Helmholtz principle [1, 2] in which both the gradient orientation and magnitude information are considered. Experimental results on a set of representative images illustrate that our proposed line segment detector, named as CannyLines, can extract more meaningful line segments than two popularly used line segment detectors, LSD [3] and ED-Lines [4], especially on the man-made scenes.

Image Orientation Estimation with Convolutional Networks

Abstract: Rectifying the orientation of scanned documents has been an important problem that was solved long ago. In this paper, we focus on the harder case of estimating and correcting the exact orientation of general images, for instance, of holiday snapshots. Especially when the horizon or other horizontal and vertical lines in the image are missing, it is hard to find features that yield the canonical orientation of the image. We demonstrate that a convolutional network can learn subtle features to predict the canonical orientation of images. In contrast to prior works that just distinguish between portrait and landscape orientation, the network regresses the exact orientation angle. The approach runs in real-time and, thus, can be applied also to live video streams.

An Efficient SIFT-Based Mode-Seeking Algorithm for Sub-Pixel Registration of Remotely Sensed Images

Abstract: Several image registration methods, based on the scaled-invariant feature transform (SIFT) technique, have appeared recently in the remote sensing literature. All of these methods attempt to overcome problems encountered by SIFT in multimodal remotely sensed imagery, in terms of the quality of its feature correspondences. The deterministic method presented in this letter exploits the fact that each SIFT feature is associated with a scale, orientation, and position to perform mode seeking (in transformation space) to eliminate outlying corresponding key points (i.e, features) and improve the overall match obtained. We also present an exhaustive empirical study on a variety of test cases, which demonstrates that our method is highly accurate and rather fast. The algorithm is capable of automatically detecting whether it succeeded or failed.

Two novel local binary pattern descriptors for texture analysis

Abstract: In this study, two novel local binary patterns were proposed.First one is based on spatial relations between neighbors with a distance parameter.The second is based on relations between a reference pixel and its neighbor on the same orientation.Two approaches are improved to detect special patterns in images.The results show that the proposed approaches can be used in image processing areas. The recent developments in the image quality, storage and data transmission capabilities increase the importance of texture analysis, which plays an important role in computer vision and image processing. Local binary pattern (LBP) is an effective statistical texture descriptor, which has successful applications in texture classification. In this paper, two novel descriptors were proposed to search different patterns in images built on LBP. One of them is based on the relations between the sequential neighbors with a specified distance and the other one is based on determining the neighbors in the same orientation through central pixel parameter. These descriptors are tested with the Brodatz-1, Brodatz-2, Butterfly and Kylberg datasets to show the applicability of the proposed nLBPd and dLBPα descriptors. The proposed methods are also compared with classical LBP. The average accuracies obtained by ANN with 10 fold cross validation, which are 99.26% (LBPu2 and nLBPd), 94.44% (dLBPα), 95.71% ( n L B P d u 2 ) and %99.64 (nLBPd), for Brodatz-1, Brodatz-2, Butterfly and Kylberg datasets, respectively, show that the proposed methods outperform significant accuracies.

Detection and estimation of ZY-3 three-line array image distortions caused by attitude oscillation

Abstract: ZY-3 is China’s first civilian high-resolution stereo surveying and mapping satellite, which is equipped with a three-line array panchromatic stereo camera. However, high-resolution satellite images (HRSIs) often suffer from satellite attitude oscillation, which can cause image distortions, and thus affect the geo-positioning accuracy. This paper presents an approach based on three-line array stereo images to detect and estimate the periodic distortions of ZY-3 that are caused by attitude oscillation. The proposed approach includes three main components, as follows: (1) A comprehensive image matching strategy, which combines the algorithms of the scale-invariant-feature-transform (SIFT), relative orientation, geometrically constrained cross-correlation (GC3), normalized cross-correlation (NCC) and least squares matching (LSM), is presented to generate dense conjugate points in multiple images. (2) A detection method is proposed to examine the relative image distortion, based on the back-projection residuals of the stereo pairs. By the use of the conjugate points, the corresponding elevation plane of each conjugate point is determined on the basis of the forward intersection. The conjugate points in an image are then forward-projected to the elevation plane to obtain their coordinates in the ground space, with the aid of the RPCs provided with the imagery. Furthermore, these ground points are back-projected to the other images. Therefore, the relative image distortions are detected by calculating the residuals between the back-projected points and the corresponding conjugate points in the image space. (3) The sum of the sinusoidal functions is presented to model the periodic distortions caused by the attitude oscillation. Based on the constructed distortion model, the absolute image distortions in the across-track direction are estimated by the use of the steepest descent algorithm. Three experiments were conducted to assess the proposed method for estimation of the periodic distortions of ZY-3 satellite images. The experimental results demonstrated the following: (1) an absolute periodic distortion of around 0.67 Hz in three-line array images of ZY-3 was, for the first time, estimated in the across-track direction. An amplitude of about 2.63 pixels was detected for the images acquired in the early period after satellite launch. However, the amplitude became smaller in the images acquired in the later periods. (2) With respect to the estimated image distortions from the intra-track images, the back-projected residuals between the inter-track images could be improved to a sub-pixel accuracy level after distortion compensation. (3) After conducting the periodic distortion compensation, the discrepancies of the check points (CKPs) could be reduced from 1.86, 1.42 and 0.73 pixels to 0.62, 0.41 and 0.51 pixels for the nadir, forward and backward images with five ground control points (GCPs), respectively.

A New Technique for Multi-Oriented Scene Text Line Detection and Tracking in Video

Abstract: Text detection and tracking in video is challenging due to contrast, resolution and background variations, and different orientations and text movements. In addition, the presence of both caption and scene texts in video aggravates the problem because these two text types differ in characteristics significantly . This paper proposes a new technique for detecting and tracking video texts of any orientation by using spatial and temporal information, respectively. The technique explores gradient directional symmetry at component level for smoothing edge components before text detection. Spatial information is preserved by forming Delaunay triangulation in a novel way at this level, which results in text candidates. Text characteristics are then proposed in a different way for eliminating false text candidates , which results in potential text candidates. Then grouping is proposed for combining potential text candidates regardless of orientation based on the nearest neighbor criterion. To tackle the problems of multi-font and multi-sized texts, we propose multi-scale integration by a pyramid structure, which helps in extracting full text lines. Then, the detected text lines are tracked in video by matching the subgraphs of triangulation. Experimental results for text detection and tracking on our video dataset, the benchmark video datasets, and the natural scene image benchmark datasets show that the proposed method is superior to the state-of-the-art methods in terms of recall, precision , and F-measure.

A Probabilistic Framework for Joint Pedestrian Head and Body Orientation Estimation

Abstract: We present a probabilistic framework for the joint estimation of pedestrian head and body orientation from a mobile stereo vision platform. For both head and body parts, we convert the responses of a set of orientation-specific detectors into a (continuous) probability density function. The parts are localized by means of a pictorial structure approach, which balances part-based detector responses with spatial constraints. Head and body orientations are estimated jointly to account for anatomical constraints. The joint single-frame orientation estimates are integrated over time by particle filtering. The experiments involved data from a vehicle-mounted stereo vision camera in a realistic traffic setting; 65 pedestrian tracks were supplied by a state-of-the-art pedestrian tracker. We show that the proposed joint probabilistic orientation estimation framework reduces the mean absolute head and body orientation error up to 15° compared with simpler methods. This results in a mean absolute head/body orientation error of about 21°/19°, which remains fairly constant up to a distance of 25 m. Our system currently runs in near real time (8–9 Hz).

Identity verification using shape and geometry of human hands

Abstract: Shape and geometry features are encoded from contour of the hand only.Robust preprocessing is introduced to cope with the noise and disjoint fingers.Hand orientation and finger registration is applied to provide more flexibility.Two level score fusion is adopted to enhance the verification performance.Promising results are obtained over contact and contactless (IITD) datasets. A multimodal biometric system for personal identity verification is proposed using hand shape and hand geometry in this paper. Shape and geometry features are derived with the help of only contour of the hand image for which only one image acquisition device is sufficient. All the processing is done with respect to a stable reference point at wrist line which is more stable as compared to the centroid against the finger rotation and peaks and valleys determination. Two shape based features are extracted by using the distance and orientation of each point of hand contour with respect to the reference point followed by wavelet decomposition to reduce the dimension. Seven distances are used to encode the geometrical information of the hand. Shape and geometry based features are fused at score levels and their performances are evaluated using standard ROC curves between false acceptance rate, true acceptance rate, equal error rate and decidability index. Different similarity measures are used to examine the accuracy of the introduced method. Performance of system is analyzed for shape based (distance and orientation) and geometrical features individually as well as for all possible combinations of feature and score level fusion. The proposed features and fusion methods are studied over two hand image datasets, (1) JUET contact database of 50 subjects having 10 templates each and (2) IITD contactless dataset of 240 subjects with 5 templates each. The proposed method outperforms other approaches with the best 0.31% of EER.

Automated Detection of Rockfish in Unconstrained Underwater Videos Using Haar Cascades

Abstract: This article presents methods applied for automated detection of fish based on cascade classifiers of Haar-like features created using underwater images from a remotely operated vehicle under ocean survey conditions The images are unconstrained, and the imaging environment is highly variable due to the moving imaging platform, a complex rocky seabed background, and still and moving cryptic fish targets These images are released in a new image dataset, "labeled fishes in the wild," of in situ groundfishes, mainly rockfishes (Sebastes spp) and other associated species The dataset includes an annotated training and validation image set, as well as an independent test video image sequence Several Haar cascades are developed from the training set and applied to the validation and test video images for evaluation Based on performance evaluation using the validation set, true positive detection rates of 63 to 89% were achieved for seven classifiers True positive detection rates for the test video were 66% to 81% for analyst-confirmed fish targets Detector performance is dependent on training data, training and detection parameters, fish orientation, range to target, variable light intensity and attenuation

Adapted Anisotropic Gaussian SIFT Matching Strategy for SAR Registration

Abstract: In this letter, we propose an adapted anisotropic Gaussian scale-invariant feature transform (AAG-SIFT) method to find feature matches for synthetic aperture radar (SAR) image registration. First, features are detected and described in an AAG scale space. The scale space is built adaptively to local structures. Noises are blurred, but details and edges remain unaffected in this scale space. Compared with traditional SIFT-based matching methods, features extracted by AAG-SIFT are more stable and precise. Then, the dominant orientation consistency (DOC) property is analyzed and adopted to improve the matching stability. The correct matching rate is significantly increased by DOC matching. Experiments on various SAR images demonstrate the applicability of AAG-SIFT to find stable and precise feature matches for SAR registration.