Bounding Rectangle Research Articles

Visual positioning technology of picking robots for dynamic litchi clusters with disturbance

Visual localization by picking robots in natural environments is a difficult problem due to disturbance by interference factors. To solve this problem, research on visual positioning for dynamic litchi clusters with disturbance under natural environments was performed in this study. A visual system with two CCD cameras was used to acquire images of dynamic litchi clusters. The centroid of minimum bounding rectangle of litchi fruit was calculated, and the pendulum principle was used to calculate the oscillation angle of litchi clusters with three disturbance states: static, slight and large. For static or slight disturbance, the improved fuzzy C-means clustering method was used in image segmentation to obtain the litchi fruit and stem, and picking points were calculated using binocular visual stereo matching. The indoor experimental results show the maximum depth error of the picking point visual positioning for litchi clusters with static or slight disturbance is 5.08 cm, and the minimum depth error is 1.96 cm. The orchard visual positioning test results show the maximum depth error is 5.8 cm, and the minimum depth error is 0.4 cm. The results meet the picking demands of the picking robot end effector.

Spatiotemporal Interpolation Methods for Solar Event Trajectories

Abstract This paper introduces four spatiotemporal interpolation methods that enrich complex, evolving region trajectories that are reported from a variety of ground-based and space-based solar observatories every day. Our interpolation module takes an existing solar event trajectory as its input and generates an enriched trajectory with any number of additional time–geometry pairs created by the most appropriate method. To this end, we designed four different interpolation techniques: MBR-Interpolation (Minimum Bounding Rectangle Interpolation), CP-Interpolation (Complex Polygon Interpolation), FI-Interpolation (Filament Polygon Interpolation), and Areal-Interpolation, which are presented here in detail. These techniques leverage k-means clustering, centroid shape signature representation, dynamic time warping, linear interpolation, and shape buffering to generate the additional polygons of an enriched trajectory. Using ground-truth objects, interpolation effectiveness is evaluated through a variety of measures based on several important characteristics that include spatial distance, area overlap, and shape (boundary) similarity. To our knowledge, this is the first research effort of this kind that attempts to address the broad problem of spatiotemporal interpolation of solar event trajectories. We conclude with a brief outline of future research directions and opportunities for related work in this area.

Copy-move forgery detection of duplicated objects using accurate PCET moments and morphological operators

ABSTRACTThis paper presents a new method for copy-move forgery detection of duplicated objects. A bounding rectangle is drawn around the detected object to form a sub-image. Morphological operator is used to remove the unnecessary small objects. Highly accurate polar complex exponential transform moments are used as features for the detected objects. Euclidian distance and correlation coefficient between the feature vectors are calculated and used for searching the similar objects. A set of 20 forged images with duplicated objects is carefully selected from previously published works. Additional 80 non-forged images are edited by the authors and forged by duplicating different kinds of objects. Numerical simulation is performed where the results show that the proposed method successfully detect different kinds of duplicated objects. The proposed method is much faster than the previously existing methods. Also, it exhibits high robustness to various attacks such as additive white Gaussian noise, JPEG compression, rotation, and scaling.

Robust Spacecraft Component Detection in Point Clouds.

Automatic component detection of spacecraft can assist in on-orbit operation and space situational awareness. Spacecraft are generally composed of solar panels and cuboidal or cylindrical modules. These components can be simply represented by geometric primitives like plane, cuboid and cylinder. Based on this prior, we propose a robust automatic detection scheme to automatically detect such basic components of spacecraft in three-dimensional (3D) point clouds. In the proposed scheme, cylinders are first detected in the iteration of the energy-based geometric model fitting and cylinder parameter estimation. Then, planes are detected by Hough transform and further described as bounded patches with their minimum bounding rectangles. Finally, the cuboids are detected with pair-wise geometry relations from the detected patches. After successive detection of cylinders, planar patches and cuboids, a mid-level geometry representation of the spacecraft can be delivered. We tested the proposed component detection scheme on spacecraft 3D point clouds synthesized by computer-aided design (CAD) models and those recovered by image-based reconstruction, respectively. Experimental results illustrate that the proposed scheme can detect the basic geometric components effectively and has fine robustness against noise and point distribution density.

On the spatial distribution of buildings for map generalization

ABSTRACTInformation on spatial distribution of buildings must be explored as part of the process of map generalization. A new approach is proposed in this article, which combines building classification and clustering to enable the detection of class differences within a pattern, as well as patterns within a class. To do this, an analysis of existing parameters describing building characteristics is performed via principal component analysis (PCA), and four major parameters (i.e. convex hull area, IPQ compactness, number of edges, and smallest minimum bounding rectangle orientation) are selected for further classification based on similarities between building characteristics. A building clustering method based on minimum spanning tree (MST) considering rivers and roads is then applied. Theory and experiments show that use of a relative neighbor graph (RNG) is more effective in detecting linear building patterns than either a nearest neighbor graph (NNG), an MST, or a Gabriel graph (GssG). Building classification and clustering are therefore conducted separately using experimental data extracted from OpenStreetMap (OSM), and linear patterns are then recognized within resultant clusters. Experimental results show that the approach proposed in this article is both reasonable and efficient for mining information on the spatial distribution of buildings for map generalization.

Spatial Index for Uncertain Time Series

A search for patterns in uncertain time series is time-expensive in today's large databases using the currently available methods. To accelerate the search process for uncertain time series data, in this paper, we explore a spatial index structure, which uses uncertain information stored in minimum bounding rectangle and ameliorates the general prune/search process along the path from the root to leaves. To get a better performance, we normalize the uncertain time series using the weighted variance before the prune/hit process. Meanwhile, we add two goodness measures with respect to the variance to improve the robustness. The extensive experiments show that, compared with the primitive probabilistic similarity search algorithm, the prune/hit process of the spatial index can be more efficient and robust using the specific preprocess and variant index operations with just a little loss of accuracy.

Detection of Maize Navigation Centerline Based on Machine Vision

For the problem of automatic mechanical navigation in maize fields, we studied the maize navigation centerline by collecting maize root images. In this paper, through the RGB color model, the red feature is used to extract the roots of the maize plants. Next, according to the shape characteristics of the root target, the minimum bounding rectangle is used to determine the positioning character points of maize roots. Then, the least square method is used to fit these positioning points which are obtained by calculation in order to extract the root row lines. Finally, by calculating the slopes of the root row lines, we are able to get the actual navigation centerline. The experiment showed the result that compared with other algorithms, the algorithm of this paper takes less time to extract the navigation centerline. In other different environments, the accuracy of the navigation centerline extracted by this algorithm is above 92%. Therefore, this algorithm has strong robustness and real-time performance. This paper provides a reliable navigation method for the autonomous walking of agricultural machinery in maize fields.

Position Detection and Direction Prediction for Arbitrary-Oriented Ships via Multitask Rotation Region Convolutional Neural Network

Ship detection is of great importance and full of challenges in the field of remote sensing. The complexity of application scenarios, the redundancy of detection region, and the difficulty of dense ship detection are all the main obstacles that limit the successful operation of traditional methods in ship detection. In this paper, we propose a brand new detection model based on multitask rotational region convolutional neural network to solve the problems above. This model is mainly consist of five consecutive parts: Dense Feature Pyramid Network (DFPN), adaptive region of interest (ROI) Align, rotational bounding box regression, prow direction prediction and rotational nonmaximum suppression (R-NMS). First of all, the low-level location information and high-level semantic information are fully utilized through multiscale feature networks. Then, we design Adaptive ROI Align to obtain high quality proposals which remain complete spatial and semantic information. Unlike most previous approaches, the prediction obtained by our method is the minimum bounding rectangle of the object with less redundant regions. Therefore, rotational region detection framework is more suitable to detect the dense object than traditional detection model. Additionally, we can find the berthing and sailing direction of ship through prediction. A detailed evaluation based on SRSS for rotation detection shows that our detection method has a competitive performance.

Automatic Detection of Acromegaly From Facial Photographs Using Machine Learning Methods

BackgroundAutomatic early detection of acromegaly is theoretically possible from facial photographs, which can lessen the prevalence and increase the cure probability. MethodsIn this study, several popular machine learning algorithms were used to train a retrospective development dataset consisting of 527 acromegaly patients and 596 normal subjects. We firstly used OpenCV to detect the face bounding rectangle box, and then cropped and resized it to the same pixel dimensions. From the detected faces, locations of facial landmarks which were the potential clinical indicators were extracted. Frontalization was then adopted to synthesize frontal facing views to improve the performance. Several popular machine learning methods including LM, KNN, SVM, RT, CNN, and EM were used to automatically identify acromegaly from the detected facial photographs, extracted facial landmarks, and synthesized frontal faces. The trained models were evaluated using a separate dataset, of which half were diagnosed as acromegaly by growth hormone suppression test. ResultsThe best result of our proposed methods showed a PPV of 96%, a NPV of 95%, a sensitivity of 96% and a specificity of 96%. ConclusionsArtificial intelligence can automatically early detect acromegaly with a high sensitivity and specificity.

Task-Management Method Using R-Tree Spatial Cloaking for Large-Scale Crowdsourcing

With the development of sensor technology and the popularization of the data-driven service paradigm, spatial crowdsourcing systems have become an important way of collecting map-based location data. However, large-scale task management and location privacy are important factors for participants in spatial crowdsourcing. In this paper, we propose the use of an R-tree spatial cloaking-based task-assignment method for large-scale spatial crowdsourcing. We use an estimated R-tree based on the requested crowdsourcing tasks to reduce the crowdsourcing server-side inserting cost and enable the scalability. By using Minimum Bounding Rectangle (MBR)-based spatial anonymous data without exact position data, this method preserves the location privacy of participants in a simple way. In our experiment, we showed that our proposed method is faster than the current method, and is very efficient when the scale is increased.

Image-based automated potato tuber shape evaluation

Potato tuber length to width (L/W) ratio is an indicator of shape phenotype, which is an important quality trait assessed in breeding and variety development. The standard method of measurement using calipers is labor intensive and time consuming. In this study, an image acquisition system was integrated with an automated potato sizer to capture video data of tubers during sorting for estimation of L/W ratios. An algorithm was developed to segment and estimate the L/W ratios from the video frame in an accurate and high-throughput manner. Line profile was used to determine the tuber position in the frame. The minimal bounding rectangle of each tuber was computed to estimate length and width of the tubers. The imaging conditions (light, imaging distance, and speed) were optimized using fresh market potato tubers (43 samples). Finally, the algorithm was tested with eight sets of field samples of tubers of cultivars Bondi and Alturas (about 709–1273 samples/set). Optimization results indicated that L/W measurement accuracy was higher than 95% for the fresh market potato tubers, with no significant effect of tested imaging conditions. There was also a significant correlation between ground-truth caliper measurements and image-based data (Pearson’s correlation coefficient: 0.84–0.99, p < 0.01). The accuracies of L/W estimations for field samples of Bondi and Alturas tubers ranged from 76 to 100%. The lower accuracies are likely attributed to differences in sample size. Nevertheless, the method is applicable for rapid and accurate estimation of L/W ratio for a large set of samples .

An Optimization Approach for Localization Refinement of Candidate Traffic Signs

We propose a localization refinement approach for candidate traffic signs. Previous traffic sign localization approaches, which place a bounding rectangle around the sign, do not always give a compact bounding box, making the subsequent classification task more difficult. We formulate localization as a segmentation problem, and incorporate prior knowledge concerning color and shape of traffic signs. To evaluate the effectiveness of our approach, we use it as an intermediate step between a standard traffic sign localizer and a classifier. Our experiments use the well-known German Traffic Sign Detection Benchmark (GTSDB) as well as our new Chinese Traffic Sign Detection Benchmark. This newly created benchmark is publicly available, 1 and goes beyond previous benchmark data sets: it has over 5000 high-resolution images containing more than 14 000 traffic signs taken in realistic driving conditions. Experimental results show that our localization approach significantly improves bounding boxes when compared with a standard localizer, thereby allowing a standard traffic sign classifier to generate more accurate classification results. 1 http://cg.cs.tsinghua.edu.cn/ctsdb/

Efficient extraction of spatial relations for extended objects vis-à-vis human activity recognition in video

Human activity recognition (HAR) deals with recognition of activities or interactions that include humans within a video. Entities occurring in a video frame can be abstracted in variety of ways, ranging from the detailed silhouette of the entity to the very basic axis-aligned minimum bounding rectangles (MBR). On one end of the spectrum, using a detailed silhouette is not only demanding in terms of storage and computational resources but is also easily susceptible to noise. On the other end of the spectrum, MBRs require less storage, computation and abstracts out noise and video specific details. However, for abstraction of human bodies in a video an MBR does not offer an adequate solution because in addition to abstracting away noise, it also abstracts out important details such as the posture of the human body. For a more precise description, which offers a reasonable tradeoff between efficiency and noise elimination, a human body can be abstracted using a set of MBRs corresponding to different body parts. However, for a representation of activities as relations between interacting objects, a simplistic approximation assuming each MBR to be an independent entity leads to computation of redundant relations. In this paper, we explore a representation schema for interaction between entities that are considered as sets of rectangles, also referred to as extended objects. We further show that, given the representation schema, a simple recursive algorithm can be used to opportunistically extract topological, directional and distance information in O(n l o g n) time. We evaluate our representation schema for HAR on the Mind’s Eye dataset ( http://www.visint.org ), the UT-Interaction (Ryoo and Aggarwal 2010) dataset and the SBU Kinect Interaction dataset (Yun et al. 2012).

Traffic flow detection and statistics via improved optical flow and connected region analysis

Moving vehicle detection plays an important role in intelligent transportation systems. One of the common methods used in moving vehicle detection is optical flow. However, conventional Horn–Schunck optical flow consumes too much time when calculating dense optical flows so that it cannot meet the real-time requirements. This paper proposes a novel improved Horn–Schunck optical flow algorithm based on inter-frame differential method. In our algorithm, optical flow field distribution is only calculated for pixels with larger gray values in the difference image, while for other pixels we applied the iterative smooth. The number of vehicles in the videos of traffic conditions is counted by setting the virtual loop and detecting optical flow information. To extract the moving vehicle as accurately as possible, we also propose a method to obtain moving vehicle minimum bounding rectangle based on the connected region analysis. Finally, we compare the improved optical flow with other four optical flow algorithms in moving vehicle extraction and vehicle flow detection, from which our method gives a much more accurate result.

Simulation study of the impact of local real-time traffic information provision strategy in connected vehicle systems

Abstract Current vehicle navigation systems usually rely on global real-time traffic information. They suggest the best paths to travelers through an independent and selfish-routing mechanism. Existing studies show that such systems have a great potential to cause traffic congestions oscillating among several alternative corridors, i.e., iteratively leading exceeding traffic into light traffic corridors and making congestion on one and then another within a short time period. To address this issue, this research proposes a local information provision strategy, given travelers are still guided by an independent and selfish-routing mechanism. The coverage of the real-time information provision is identified by the minimum bounding rectangle, and the ellipse method respectively. The benefits of the proposed strategy in system level are investigated by the simulation studies built upon the Borman expressway network. The simulation results indicate that as the local travel time information provision strategy is used, it may save 10–25% of the system travel time resulting from global information provision strategy under medium or high network traffic loads. Moreover, the sensitivity analyses indicate that the merits of the local travel time information provision strategy are more significant as the penetration of smart vehicles increases over the network. Thus, the proposed local information provision strategy outperforms global information provision strategy in system level network performance.

Image Processing of Porous Silicon Microarray in Refractive Index Change Detection.

A new method for extracting the dots is proposed by the reflected light image of porous silicon (PSi) microarray utilization in this paper. The method consists of three parts: pretreatment, tilt correction and spot segmentation. First, based on the characteristics of different components in HSV (Hue, Saturation, Value) space, a special pretreatment is proposed for the reflected light image to obtain the contour edges of the array cells in the image. Second, through the geometric relationship of the target object between the initial external rectangle and the minimum bounding rectangle (MBR), a new tilt correction algorithm based on the MBR is proposed to adjust the image. Third, based on the specific requirements of the reflected light image segmentation, the array cells are segmented into dots as large as possible and the distance between the dots is equal in the corrected image. Experimental results show that the pretreatment part of this method can effectively avoid the influence of complex background and complete the binarization processing of the image. The tilt correction algorithm has a shorter computation time, which makes it highly suitable for tilt correction of reflected light images. The segmentation algorithm makes the dots in a regular arrangement, excludes the edges and the bright spots. This method could be utilized in the fast, accurate and automatic dots extraction of the PSi microarray reflected light image.

Automatic detection of Martian dark slope streaks by machine learning using HiRISE images

Dark slope streaks (DSSs) on the Martian surface are one of the active geologic features that can be observed on Mars nowadays. The detection of DSS is a prerequisite for studying its appearance, morphology, and distribution to reveal its underlying geological mechanisms. In addition, increasingly massive amounts of Mars high resolution data are now available. Hence, an automatic detection method for locating DSSs is highly desirable. In this research, we present an automatic DSS detection method by combining interest region extraction and machine learning techniques. The interest region extraction combines gradient and regional grayscale information. Moreover, a novel recognition strategy is proposed that takes the normalized minimum bounding rectangles (MBRs) of the extracted regions to calculate the Local Binary Pattern (LBP) feature and train a DSS classifier using the Adaboost machine learning algorithm. Comparative experiments using five different feature descriptors and three different machine learning algorithms show the superiority of the proposed method. Experimental results utilizing 888 extracted region samples from 28 HiRISE images show that the overall detection accuracy of our proposed method is 92.4%, with a true positive rate of 79.1% and false positive rate of 3.7%, which in particular indicates great performance of the method at eliminating non-DSS regions.

Camera Coverage Estimation Based on Multistage Grid Subdivision

Visual coverage is one of the most important quality indexes for depicting the usability of an individual camera or camera network. It is the basis for camera network deployment, placement, coverage-enhancement, planning, etc. Precision and efficiency are critical influences on applications, especially those involving several cameras. This paper proposes a new method to efficiently estimate superior camera coverage. First, the geographic area that is covered by the camera and its minimum bounding rectangle (MBR) without considering obstacles is computed using the camera parameters. Second, the MBR is divided into grids using the initial grid size. The status of the four corners of each grid is estimated by a line of sight (LOS) algorithm. If the camera, considering obstacles, covers a corner, the status is represented by 1, otherwise by 0. Consequently, the status of a grid can be represented by a code that is a combination of 0s or 1s. If the code is not homogeneous (not four 0s or four 1s), the grid will be divided into four sub-grids until the sub-grids are divided into a specific maximum level or their codes are homogeneous. Finally, after performing the process above, total camera coverage is estimated according to the size and status of all grids. Experimental results illustrate that the proposed method’s accuracy is determined by the method that divided the coverage area into the smallest grids at the maximum level, while its efficacy is closer to the method that divided the coverage area into the initial grids. It considers both efficiency and accuracy. The initial grid size and maximum level are two critical influences on the proposed method, which can be determined by weighing efficiency and accuracy.

Shape-Based Building Detection in Visible Band Images Using Shadow Information

This paper introduces a novel methodology for automated detection of buildings from single high-resolution optical images with only visible red, green, and blue bands of data. In particular, we first investigate the shadow evidence to focus on building regions. Then, a novel Markov random field (MRF)-based region growing segmentation technique is proposed. Image is oversegmented into smaller homogeneous regions that can be used to replace the rigid structure of the pixel grid. An iterative classification merging is then applied over this set of regions. At each iteration, regions are classified using a region-level MRF model, then, according to the position of shadows, regions having the same class are merged to produce new regions whose shapes are appropriate to rectangles. The final buildings are determined using a recursive minimum bounding rectangle. The experimental results prove that the proposed method is applicable in various areas (high dense urban, suburban, and rural) and is highly robust and reliable.

SReN: Shape Regression Network for Comic Storyboard Extraction

The goal of storyboard extraction is to decompose the comic image into several storyboards(or frames), which is the fundamental step of comic image understanding and producing digital comic documents suitable for mobile reading. Most of existing approaches are based on hand crafted low-level visual patters like edge segments and line segments, which do not capture high-level vision. To overcome shortcomings of the existing approaches, we propose a novel architecture based on deep convolutional neural network, namely Shape Regression Network(SReN), to detect storyboards within comic images. Firstly, we use Fast R-CNN to generate rectangle bounding boxes as storyboard proposals. Then we train a deep neural network to predict quadrangles for these propos- als. Unlike existing object detection methods which only output rectangle bounding boxes, SReN can produce more precise quadrangle bounding boxes. Experimental results, evaluating on 7382 comic pages, demonstrate that SReN outperforms the state-of-the-art methods by more than 10% in terms of F1-score and page correction rate.