Showing papers on "Corner detection published in 2016"

Fast event-based Harris corner detection exploiting the advantages of event-driven cameras

Abstract: The detection of consistent feature points in an image is fundamental for various kinds of computer vision techniques, such as stereo matching, object recognition, target tracking and optical flow computation. This paper presents an event-based approach to the detection of corner points, which benefits from the high temporal resolution, compressed visual information and low latency provided by an asynchronous neuromorphic event-based camera. The proposed method adapts the commonly used Harris corner detector to the event-based data, in which frames are replaced by a stream of asynchronous events produced in response to local light changes at μs temporal resolution. Responding only to changes in its field of view, an event-based camera naturally enhances edges in the scene, simplifying the detection of corner features. We characterised and tested the method on both a controlled pattern and a real scenario, using the dynamic vision sensor (DVS) on the neuromorphic iCub robot. The method detects corners with a typical error distribution within 2 pixels. The error is constant for different motion velocities and directions, indicating a consistent detection across the scene and over time. We achieve a detection rate proportional to speed, higher than frame-based technique for a significant amount of motion in the scene, while also reducing the computational cost.

System and method for detecting objects in an image

Abstract: A method for cropping photos images captured by a user from an image of a page of a photo album is described. Corners in the page image are detected using corner detection algorithm or by detecting intersections of line-segments (and their extensions) in the image using edge, corner, or line detection techniques. Pairs of the detected corners are used to define all potential quads, which are then are qualified according to various criteria. A correlation matrix is generated for each potential pair of the qualified quads, and candidate quads are selected based on the Eigenvector of the correlation matrix. The content of the selected quads is checked using a salience map that may be based on a trained neuron network, and the resulting photos images are extracted as individual files for further handling or manipulation by the user.

Analysing Domain Shift Factors between Videos and Images for Object Detection

Abstract: Object detection is one of the most important challenges in computer vision. Object detectors are usually trained on bounding-boxes from still images. Recently, video has been used as an alternative source of data. Yet, for a given test domain (image or video), the performance of the detector depends on the domain it was trained on. In this paper, we examine the reasons behind this performance gap. We define and evaluate different domain shift factors: spatial location accuracy, appearance diversity, image quality and aspect distribution. We examine the impact of these factors by comparing performance before and after factoring them out. The results show that all four factors affect the performance of the detectors and their combined effect explains nearly the whole performance gap.

Unsupervised SAR Image Change Detection Based on SIFT Keypoints and Region Information

Abstract: This letter presents a new unsupervised distribution-free change detection method for synthetic aperture radar (SAR) images based on scale-invariant feature transform (SIFT) keypoints and region information. Since the SIFT can detect bloblike structures in an image and be insensitive to noise, we first extract noise-robust SIFT keypoints in the log-ratio image to reduce the detection range. Then, in order to obtain accurate changed regions, rather than directly obtaining the change-detection map from the difference image as in some traditional change detection methods, we make segmentation around the extracted keypoints in the two original multitemporal SAR images, where the edges of detection regions are much clearer than those in the difference image, and further compare the two segmentations to generate the change-detection map. This method utilizes the bloblike structure information offered by SIFT keypoints and the region information extracted via image segmentation. Experiments on real SAR images demonstrate the effectiveness of the proposed method.

A Survey of Blob Detection Algorithms for Biomedical Images

Abstract: This paper presents a survey of blob detection methods which has been applied on image processing with relation of medical images proposed by literature. “The blob detection is a mathematical method which detects regions or points in digital images”. [1] The regions or points which have noticeable difference with their surroundings is called blob. Given the increased interest in biomedical image processing system, many algorithms and methods have been reported to apply but there is no systematic survey and classification of the blob detection for medical images and how they have been assessed and applied. The findings, which is the most usable methods of blob detectors in biomedical image processing has been presented. It was also investigated how these studies have been surveyed, how they evolved in the main digital libraries over the last decade, and what points deserves further attention, through new research. From this survey, practitioners and researchers can adopt the blob detection methods and analyze to use these methods in their research for further development.

Indoor robot SLAM method and system

Abstract: The invention discloses an indoor robot SLAM method. The method comprises the following steps: obtaining image data through a camera, wherein the image data comprises RGB images and depth images; using a corner detection algorithm, an LK characteristic point tracing algorithm, and a RANSAC algorithm to process the RGB images and depth images to adjust the position and angle of the camera so as to obtain the RGB image data information under a robot operation system; converting the depth images in a world coordinate system to a ground coordinate, traversing the depth images that are projected on the ground coordinate, setting the gray value of the grid where a barrier stays as a first characteristic value, carrying out traversing to obtain a 2D barrier grid map; through a single line laser radar scanning mode, searching the 2D barrier grid map, when the grid with a grey value equal to the first characteristic value is found, feeding back the distance between the grid and the camera to obtain the distance between a robot and the barrier, obtaining the depth images under a robot operation system, and obtaining an environment 2D map.

Automatic chessboard corner detection method

Abstract: Chessboard corner detection is a necessary procedure of the popular chessboard pattern-based camera calibration technique, in which the inner corners on a two-dimensional chessboard are employed as calibration markers. In this study, an automatic chessboard corner detection algorithm is presented for camera calibration. In authors' method, an initial corner set is first obtained with an improved Hessian corner detector. Then, a novel strategy that utilises both intensity and geometry characteristics of the chessboard pattern is presented to eliminate fake corners from the initial corner set. After that, a simple yet effective approach is adopted to sort the detected corners into a meaningful order. Finally, the sub-pixel location of each corner is calculated. The proposed algorithm only requires a user input of the chessboard size, while all the other parameters can be adaptively calculated with a statistical approach. The experimental results demonstrate that the proposed method has advantages over the popular OpenCV chessboard corner detection method in terms of detection accuracy and computational efficiency. Furthermore, the effectiveness of the proposed method used for camera calibration is also verified in authors' experiments.

An improved method for fisheye camera calibration and distortion correction

Abstract: The fisheye camera has been widely studied in the field of robot vision since it can capture a wide view of the scene at one time However, serious image distortion handers it from being widely used To remedy this, this paper proposes an improved fisheye lens calibration and distortion correction method First, an improved automatic detection of checkerboards is presented to avoid the original constraint and user intervention that usually existed in the conventional methods A state-of-the-art corner detection method is evaluated and its strengths and shortcomings are analyzed An adaptively automatic corner detection algorithm is implemented to overcome the shortcomings Then, a precise mathematical model based on the law of fisheye lens imaging is modeled, which assumes that the imaging function can be described by a Taylor series expansion, followed by a nonlinear refinement based on the maximum likelihood criterion With the proposed corner detection and mathematical model of fisheye imaging, both intrinsic and external parameters of the fisheye camera can be correctly calibrated Finally, the radial distortion of the fisheye image can be corrected by incorporating the calibrated parameters Experimental results validate the effectiveness of the proposed method

Analysis of Harris corner detection for color images

Abstract: Corner detection algorithm is an algorithm which helps in identifying the corners in an image. Corners are mainly formed by the combination of two or more edges. These corners may or may not define the boundary of an image. Here the method used is Harris corner detection algorithm. It helps in pointing out the corners in a color image, for each component. This improves the detection efficiency and the experimental results shows that this method is more reliable than traditional methods.

Mining Personal Frequent Routes via Road Corner Detection

Abstract: Frequent route is an important individual outdoor behavior pattern that many trajectory-based applications rely on. In this paper, we propose a novel framework for extracting frequent routes from personal GPS trajectories. The key idea of our design is to accurately detect road corners and utilize these new metaphors to tackle the problem of frequent route extraction. Concretely, our framework contains three phases: 1) characteristic point (CP) extraction; 2) corner detection; and 3) trajectory mapping. In the first phase, we present a linear fitting-based algorithm to extract CPs. In the second phase, we develop a multiple density level DBSCAN (density-based spatial clustering of applications with noise) algorithm to locate road corners by clustering CPs. In the third phase, we convert each trajectory into an ordered sequence of road corners and obtain all routes that have been traversed by an individual for at least ${F}$ (frequency threshold) times. We evaluate the framework using real-world trajectory datasets of individuals for one year and the experimental results demonstrate that our framework outperforms the baseline approach by 7.8% on average in terms of precision and 21.9% in terms of recall.

Robust fast corner detector based on filled circle and outer ring mask

Abstract: A novel mask with a filled circle and outer ring is proposed to detect corners from images, based on the adaptive threshold of a local region. First, the inner filled circle is used in a response function to filter four non-corner regions: image noise, object edges, corner neighbourhoods, and flat regions. Second, corner candidates are detected using a complex response function, by considering the margin of inner circle and the outer ring together. Finally, related algorithms are developed to determine and remove the false corners lying on thin-band, noisy, and salient pixels. The authors' approach has been tested on artificial, noisy, fuzzy, and real images, and its performance is evaluated, analysed, and compared with the existing grey-level-based corner detection methods of Harris, SUSAN, FAST, and Lan and Zhang. The presented corner detector has better detection accuracy, less sensitivity to noisy and fuzzy images, high computational efficiency, and good repeatability.

Harris Scale Invariant Corner Detection Algorithm Based on the Significant Region

Abstract: The traditional Harris corner detection algorithm is sensitive to scale change, corners detected throughout the entire image under complex background, thus extracting more false corners, lead to the follow-up of large amount of calculation and a high rate of error matching. To solve this problem, this paper proposes an optimized Harris corner detection algorithm. First, a significant region detection method is used to extract the target area, and take closing operation for the result figure, can effectively achieve target and background segmentation; second, scale invariant describing methods is applied to Harris algorithm, at the same time, combined with the non-maximum suppression methods to extract corners, get more right corners. Through experiment contrasts, the algorithm used in this paper can be improved more corner detection performance.

Image Mosaicing Using Binary Edge Detection Algorithm in a Cloud-Computing Environment

Abstract: Image Mosaicing is an image processing technique that arises from the need of having a more realistic view of the real world wider than the view captured by the lenses of the available cameras. In this paper, a sequence of images will be mosaiced using binary edge detection algorithm in a cloud-computing environment to improve processing speed and accuracy. The authors have used Platform as a Service PaaS to provide a number of nodes in the cloud to run the computational intensive image processing and stitching algorithms. This increased the processing speed as most of image processing algorithms deal with every single pixel in the image. Message Passing Interface MPI is used for message passing among the compute-nodes in the cloud and a MapReduce technique is used for image distribution and collection, where the root node is used as reducer and the others as mappers. After applying the algorithm on different sequence of images and different machines on JUST cloud, the authors have achieved high mosaicing accuracy, and the execution time has been improved when comparing it with sequential execution on the images.

Reliably detecting humans with RGB-D camera with physical blob detector followed by learning-based filtering

Abstract: This paper proposes a two-staged approach to real-time human detection in cluttered environments using RGB-D camera. The first stage is a novel physical blob (P-Blob) detection that can quickly find plausible human heads. The second stage uses a combination of human upper-body features to filter out false positives. Experiment results on three publicly available datasets show that the proposed method can reliably detect people in RGB-D video in real time.

Enhanced low-complexity pruning for corner detection

Abstract: Low-complexity corner detection is essential for many real-time computer vision applications that need to be executed on low-cost/low-power embedded platforms such as robots. The widely used Shi---Tomasi and Harris corner detectors become prohibitive in such platforms due to their high computational complexity, which is attributed to the need to apply a complex corner measure on the entire image. In this paper, we introduce a novel and computationally efficient technique to accelerate the Shi---Tomasi and Harris corner detectors. The proposed technique consists of two steps. In the first step, the complex corner measure is replaced with simple approximations to quickly prune away non-corners. In the second step, the complex corner measure is applied to a small corner candidate set obtained after pruning. Evaluations using standard image benchmarks show that the proposed pruning technique achieves up to 75 % speedup on the Nios-II platform, while yielding corners with comparable or better accuracy than the conventional Shi---Tomasi and Harris detectors.

LEA Detection and Tracking Method for Color-Independent Visual-MIMO.

Abstract: Communication performance in the color-independent visual-multiple input multiple output (visual-MIMO) technique is deteriorated by light emitting array (LEA) detection and tracking errors in the received image because the image sensor included in the camera must be used as the receiver in the visual-MIMO system. In this paper, in order to improve detection reliability, we first set up the color-space-based region of interest (ROI) in which an LEA is likely to be placed, and then use the Harris corner detection method. Next, we use Kalman filtering for robust tracking by predicting the most probable location of the LEA when the relative position between the camera and the LEA varies. In the last step of our proposed method, the perspective projection is used to correct the distorted image, which can improve the symbol decision accuracy. Finally, through numerical simulation, we show the possibility of robust detection and tracking of the LEA, which results in a symbol error rate (SER) performance improvement.

Vehicle detection and tracking based on corner and lines adjacent detection features

Abstract: This paper discusses a new method in detecting moving objects, which is differ from most of the methods used such as Gaussian Mixture Model, and Haar-Like approach. The focus is on utilizing corner detection and line adjacent detection features through thresholding process creating black and white images to detect the corner of each object in each frame. The process divides a frame length into 4 parts, whereas the first part acted as initiation process of moving object recognition while the rest of the frame functioning as vehicle tracking, speed measurement, and number of vehicles calculation. The initiation process started by identifying corner spots of the moving objects that must be recognized as a single object. The lines surpassing through two points are later identified to determine whether those spots have dark color (0) or light color (1). The moving objects is represented by light color (1) and the walking objects is represented by dark color (0). A group of corner spots, identified and connected by two-point-line equation to be recognized as one unified object by using corner and line adjacent method. The identified vehicle objects can be more easily tracked and identified by the average speed in order to obtain the number of passing vehicles. The research result shows that in the initiation phase, the corner and line adjacent features able to detect moving object and distinguish it with different objects. Furthermore, in tracking phase, system is able to track the vehicle position, measuring the speed and number of vehicles. The system is proven to be able to recognize the moving objects quickly and accurately resulting in the more feasible process of speed measurement and tracking.

Underwater vehicle visual servo and target grasp control

Abstract: Binocular stereoscopic vision system of underwater vehicle plays an important role in promoting the exploration and development of the marine environment. In order to realize the underwater autonomous grasping task based on visual servoing, this paper studied binocular vision ranging technology of underwater vehicle and vision control technology of manipulator. By adjusting the placement and pose of the underwater calibration board, the calibration error was reduced. In terms of the characteristics of less feature points detected by the underwater images, Harris corner detection algorithm has been improved, after that, more corners are detected from underwater images. It is more difficult to accurately describe feature points of underwater images, however, by using Daisy operator to describe the key points which are detected by SIFT algorithm, not only describing the characteristics points of the image accurately, but can also indicating pixel points around the occluded area.

X-Corner Detection for Camera Calibration Using Saddle Points

Abstract: This paper discusses a corner detection algorithm for camera calibration. Calibration is a necessary step in many computer vision and image processing applications. Robust corner detection for an image of a checkerboard is required to determine intrinsic and extrinsic parameters. In this paper, an algorithm for fully automatic and robust X-corner detection is presented. Checkerboard corner points are automatically found in each image without user interaction or any prior information regarding the number of rows or columns. The approach represents each X-corner with a quadratic fitting function. Using the fact that the X-corners are saddle points, the coefficients in the fitting function are used to identify each corner location. The automation of this process greatly simplifies calibration. Our method is robust against noise and different camera orientations. Experimental analysis shows the accuracy of our method using actual images acquired at different camera locations and orientations. Keywords—Camera Calibration, Corner Detector, Saddle Points, X-Corners.

Enhanced Breast Cancer Classification with Automatic Thresholding Using SVM and Harris Corner Detection

Abstract: Image classification and extracting the characteristics of a tumor are the powerful tools in medical science. In case of breast cancer medical treatment, the breast cancer classification methods can be used to classify input images as normal and abnormal classes for better diagnoses and earlier detection with breast tumors. However, classification process can be challenging because of the existence of noise in the images, and complicated structures of the image. Manual classification of the images is time-consuming, and need to be done only by medical experts. Hence using an automated medical image classification tool is useful and necessary. In addition, having a better training data set directly affect the quality of classification process. In this paper, a method is proposed based on supervised learning and automatic thresholding for both generating better training data set, and more accurate classification of the mammogram images into Normal/Abnormal classes. The procedure consists of preprocessing, removing noise, elimination of unwanted objects, features extraction, and classification. A Support Vector Machine (SVM) is used as the supervised model in two phases which are testing and training. Intensity value, auto-correlation matrix value of detected corners, and, energy, are three extracted features used to train the SVM. Experimental results show this method classify images with more accuracy and less execution time compared to existing method.

Stereo image feature matching using Harris corner detection algorithm

Abstract: Stereo image matching is one of the research areas in computer vision. In stereo image matching, technological developments advances from area based matching techniques to the feature based matching techniques. In this paper, we present a Harris corner detection algorithm for stereo image feature matching. This is an intensity based feature matching algorithm and it controls the strong and weak corners with the help of threshold value. Generally image processing algorithms are simulated in software but to do hardware co-simulation, here the model based design is implemented in Xilinx System Generator. Further the architecture is synthesized on the Xilinx Virtex - 5 FPGA. Simulation results are included in this paper to verify the performance of proposed system. Complexity level is minimized in model based design than that of script level design.

Object detection method and object detection apparatus

Abstract: An object detection method and an object detection apparatus are provided. The object detection method comprises: mapping at least one image frame in an image sequence into a three dimensional physical space to obtain three dimensional coordinates of each pixel in the at least one image frame; extracting a foreground region in the at least one image frame; segmenting the foreground region into a set of blobs; and detecting, for each blob in the set of blobs, an object in the blob through a neural network based on the three dimensional coordinates of at least one predetermined reference point in the blob, to obtain an object detection result.

Improved Real-Time Scan Matching Using Corner Features

Abstract: The automation of unmanned vehicle operation has gained a lot of research attention, in the last few years, because of its numerous applications. The vehicle localization is more challenging in indoor environments where absolute positioning measurements (e.g. GPS) are typically unavailable. Laser range finders are among the most widely used sensors that help the unmanned vehicles to localize themselves in indoor environments. Typically, automatic real-time matching of the successive scans is performed either explicitly or implicitly by any localization approach that utilizes laser range finders. Many accustomed approaches such as Iterative Closest Point (ICP), Iterative Matching Range Point (IMRP), Iterative Dual Correspondence (IDC), and Polar Scan Matching (PSM) handles the scan matching problem in an iterative fashion which significantly affects the time consumption. Furthermore, the solution convergence is not guaranteed especially in cases of sharp maneuvers or fast movement. This paper proposes an automated real-time scan matching algorithm where the matching process is initialized using the detected corners. This initialization step aims to increase the convergence probability and to limit the number of iterations needed to reach convergence. The corner detection is preceded by line extraction from the laser scans. To evaluate the probability of line availability in indoor environments, various data sets, offered by different research groups, have been tested and the mean numbers of extracted lines per scan for these data sets are ranging from 4.10 to 8.86 lines of more than 7 points. The set of all intersections between extracted lines are detected as corners regardless of the physical intersection of these line segments in the scan. To account for the uncertainties of the detected corners, the covariance of the corners is estimated using the extracted lines variances. The detected corners are used to estimate the transformation parameters between the successive scan using least squares. These estimated transformation parameters are used to calculate an adjusted initialization for scan matching process. The presented method can be employed solely to match the successive scans and also can be used to aid other accustomed iterative methods to achieve more effective and faster converge. The performance and time consumption of the proposed approach is compared with ICP algorithm alone without initialization in different scenarios such as static period, fast straight movement, and sharp manoeuvers.

FPGA-based real-time panoramic image mosaic method

Abstract: The invention provides an FPGA-based real-time panoramic image mosaic method. The method is characterized by, to begin with, carrying out gray conversion on collected video data; then, extracting feature points of each frame image in a video stream through a Harris corner detection method; after obtaining corner coordinates of the front-back two frame images, carrying out calculation on regions with corners as centers respectively through an NCC-SSDA fusion algorithm to find out the correspondence relation between the feature points of the two frame images; removing unmatched point pairs through an RANSAC algorithm, estimating affine transform model parameters, finding an optimum model and taking all of local inner points according with the model as the final matching result; and finally, carrying out fusion on image edges through a weighted smoothing algorithm, and seamlessly displaying images after mosaic. The method can quickly finish single-camera 360-degree panoramic mosaic, and has the advantages of high registration accuracy, high real-time performance and good stability.

Research on multi-feature front vehicle detection algorithm based on video image

Abstract: In order to meet the requirements of real time and effectiveness, a multi-feature front vehicle detection algorithm based on video image is proposed. Firstly, the proposed algorithm uses the edge detector algorithm based on wavelet transform to obtain the road mark for reducing the detection region. Secondly, the threshold segmentation algorithm based on weight factor is used for image segmentation. Thirdly, image processing technologies are used to improve the quality of the segmented image. Lastly, vehicle shadow characteristics, vertical edge characteristics, texture characteristics and symmetry characteristics are used to find out the real region of interest. Simulation demonstrates that the proposed algorithm has low mistake rate and miss rate, high detect rate and strong real-time.

Corner detection and sorting method based on improved Harris algorithm in camera calibration

Abstract: In traditional Harris corner detection algorithm, the appropriate threshold which is used to eliminate false corners is selected manually. In order to detect corners automatically, an improved algorithm which combines Harris and circular boundary theory of corners is proposed in this paper. After detecting accurate corner coordinates by using Harris algorithm and Forstner algorithm, false corners within chessboard pattern of the calibration plate can be eliminated automatically by using circular boundary theory. Moreover, a corner sorting method based on an improved calibration plate is proposed to eliminate false background corners and sort remaining corners in order. Experiment results show that the proposed algorithms can eliminate all false corners and sort remaining corners correctly and automatically.

Performance evaluation of object detection techniques for object detection

Abstract: Object detection plays vital role in image processing for finding the objects of interest Increase of image size and complexity has thrust for developing novel and robust object detection techniques. There are number of methods existing for detecting the objects in a particular scene. The focus of this paper is estimate the performance and effieciency of some existing object detection algorithms on the sampled images. Thus may give good initiative about the suitability of those algorithms on the images. Three existing algorithms are implemented for various images and compared under variety of situations to find which detector is robust under different conditions.