scispace - formally typeset
Search or ask a question

Showing papers on "Corner detection published in 2015"


Proceedings ArticleDOI
07 Jun 2015
TL;DR: This work claims that recognizing objects and predicting contours are two mutually related tasks, and shows that it can invert the commonly established pipeline: instead of detecting contours with low-level cues for a higher-level recognition task, it exploits object-related features as high- level cues for contour detection.
Abstract: Contour detection has been a fundamental component in many image segmentation and object detection systems. Most previous work utilizes low-level features such as texture or saliency to detect contours and then use them as cues for a higher-level task such as object detection. However, we claim that recognizing objects and predicting contours are two mutually related tasks. Contrary to traditional approaches, we show that we can invert the commonly established pipeline: instead of detecting contours with low-level cues for a higher-level recognition task, we exploit object-related features as high-level cues for contour detection.

354 citations


Journal ArticleDOI
TL;DR: This paper introduces an event-based luminance-free method to detect and match corner events from the output ofynchronously generating "spiking" events that encode relative changes in pixels' illumination at high temporal resolutions.

107 citations


Proceedings ArticleDOI
16 Apr 2015
TL;DR: A brief classification of the classical approaches for moving object detection is provided and recent research trends to detect moving object for single stationary camera is reviewed with discussion of key points and limitations of each approach.
Abstract: Moving object detection is the task of identifying the physical movement of an object in a given region or area. Over last few years, moving object detection has received much of attraction due to its wide range of applications like video surveillance, human motion analysis, robot navigation, event detection, anomaly detection, video conferencing, traffic analysis and security. In addition, moving object detection is very consequential and efficacious research topic in field of computer vision and video processing since it forms a critical step for many complex processes like video object classification and video tracking activity. Consequently, identification of actual shape of moving object from a given sequence of video frames becomes pertinent. However, task of detecting actual shape of object in motion becomes tricky due to various challenges like dynamic scene changes, illumination variations, presence of shadow, camouflage and bootstrapping problem. To reduce the effect of these problems, researchers have proposed number of new approaches. This paper provides a brief classification of the classical approaches for moving object detection. Further, paper reviews recent research trends to detect moving object for single stationary camera along with discussion of key points and limitations of each approach.

97 citations


Journal ArticleDOI
TL;DR: A contour-based corner detector using the angle difference of the principal directions of anisotropic Gaussian directional derivatives (ANDDs) on contours that behaves better in detection, localization, repeatability, and noise robustness is presented.

56 citations


Journal ArticleDOI
TL;DR: This work proposes two effective and efficient corner detectors using simple triangular theory and distance calculation that outperform CPDA and nine other existing corner detectors in terms of repeatability and have a relatively low or comparable localization error.

51 citations


Journal ArticleDOI
TL;DR: This paper provides an extensive experimental assessment on benchmark data sets which empirically confirms the potential of shearlets feature detection.
Abstract: Shearlets are a relatively new and very effective multi-scale framework for signal analysis. Contrary to the traditional wavelets, shearlets are capable to efficiently capture the anisotropic information in multivariate problem classes. Therefore, shearlets can be seen as the valid choice for multi-scale analysis and detection of directional sensitive visual features like edges and corners. In this paper, we start by reviewing the main properties of shearlets that are important for edge and corner detection. Then, we study algorithms for multi-scale edge and corner detection based on the shearlet representation. We provide an extensive experimental assessment on benchmark data sets which empirically confirms the potential of shearlets feature detection.

50 citations


Journal ArticleDOI
TL;DR: Second order anisotropic Gaussian kernels, which have proven successful in edge and corner detection, offer interesting advantages over isotropic kernels in ridge detection, which are illustrated on synthetic images and performed on a new dataset for line detection.

43 citations


Patent
24 Jun 2015
TL;DR: In this article, a PTAM improvement method based on ground characteristics of an intelligent robot is proposed, which includes the steps that firstly, parameter correction is completed, wherein parameter correction includes parameter definition and camera correction; secondly, current environment texture information is obtained by means of a camera, a four-layer Gausses image pyramid is constructed, the characteristic information in a current image is extracted by using FAST corner detection algorithm, data relevance between corner characteristics is established, and then a pose estimation model is obtained.
Abstract: The invention discloses a PTAM improvement method based on ground characteristics of an intelligent robot. The PTAM improvement method based on ground characteristics of the intelligent robot comprises the steps that firstly, parameter correction is completed, wherein parameter correction includes parameter definition and camera correction; secondly, current environment texture information is obtained by means of a camera, a four-layer Gausses image pyramid is constructed, the characteristic information in a current image is extracted by means of the FAST corner detection algorithm, data relevance between corner characteristics is established, and then a pose estimation model is obtained; two key frames are obtained so as to erect the camera on the mobile robot at the initial map drawing stage; the mobile robot begins to move in the initializing process, corner information in the current scene is captured through the camera and association is established at the same time; after a three-dimensional sparse map is initialized, the key frames are updated, the sub-pixel precision mapping relation between characteristic points is established by means of an extreme line searching and block matching method, and accurate re-positioning of the camera is achieved based on the pose estimation model; finally, matched points are projected in the space, so that a three-dimensional map for the current overall environment is established.

37 citations


Journal ArticleDOI
TL;DR: The main objective of this paper is to introduce a high-quality image stitching system with least computation time and concludes that ORB algorithm is the fastest, more accurate, and with higher performance.
Abstract: The construction of a high-resolution panoramic image from a sequence of input overlapping images of the same scene is called image stitching/mosaicing. It is considered as an important, challenging topic in computer vision, multimedia, and computer graphics. The quality of the mosaic image and the time cost are the two primary parameters for measuring the stitching performance. Therefore, the main objective of this paper is to introduce a high-quality image stitching system with least computation time. First, we compare many different features detectors. We test Harris corner detector, SIFT, SURF, FAST, GoodFeaturesToTrack, MSER, and ORB techniques to measure the detection rate of the corrected keypoints and processing time. Second, we manipulate the implementation of different common categories of image blending methods to increase the quality of the stitching process. From experimental results, we conclude that ORB algorithm is the fastest, more accurate, and with higher performance. In addition, Exposure Compensation is the highest stitching quality blending method. Finally, we have generated an image stitching system based on ORB using Exposure Compensation blending method.

35 citations


Patent
20 Apr 2015
TL;DR: In this article, a method, apparatus and a system multi-camera image processing method was proposed to perform geometric alignment to produce a geometric output by estimating fish eye distortion correction parameters, performing initial perspective correction on related frame, running corner detection in the overlapping areas, locating the stronger corner, calculating BRIEF descriptors for features and match feature point from two cameras using briEF scores, performing checks and rejecting wrong feature matches.
Abstract: A method, apparatus and a system multi-camera image processing method. The method includes performing geometric alignment to produce a geometric output by estimating fish eye distortion correction parameters, performing initial perspective correction on related frame, running corner detection in the overlapping areas, locating the stronger corner, calculating BRIEF descriptors for features and match feature point from two cameras using BRIEF scores, performing checks and rejecting wrong feature matches, finding perspective matrices to minimize distance between matched features; and creating a geometric lookup table.

29 citations


Proceedings ArticleDOI
01 Sep 2015
TL;DR: The proposed algorithm, called Canny Smart Routing (CannySR), runs Canny to obtain a binary edge map, and uses the Canny edgels as anchors for SR to convert them to edge segments, which can be used in many applications such as line, arc, circle, ellipse, corner detection and other similar higher level object detection applications.
Abstract: Canny Edge Detector is the most widely used operator for edge detection. The problem with Canny is that it outputs a binary edge map, where an edge pixel (edgel) is marked (e.g., its value in the edge map is 255) and a non-edge pixel is unmarked (e.g., its value in the edge map is 0). A typical binary edge map is of low quality, consisting of gaps, notch-like structures, ragged and multi-pixel wide edgels. To clean up Canny's binary edge maps, fill up one pixel-wide gaps between the edgels, and to return the map as a set of edge segments, each of which is a one-pixel wide, contiguous chain of pixels, we employ the Smart Routing (SR) algorithm from our recently proposed Edge Segment Detection Algorithm, the Edge Drawing (ED). The proposed algorithm, called Canny Smart Routing (CannySR), runs Canny to obtain a binary edge map, and uses the Canny edgels as anchors for SR to convert them to edge segments. The produced edge segments can then be used in many applications such as line, arc, circle, ellipse, corner detection and other similar higher level object detection applications. We qualitatively evaluate the effectiveness of the proposed algorithm on some sample images and conclude that CannySR visibly improves the modal quality of Canny's binary edge maps although ED seems to produce the best results.

Journal ArticleDOI
TL;DR: This paper analyzes the scale-space behavior with the Laplacian of Gaussian (LoG) operator on a planar curve which constructs LaPLacian Scale Space (LSS), and the analytical expression of a LaplACian Scale-Space map (L SS map) is obtained, demonstrating the LaPlacian scale-Space behavior of the planar Curve corners.
Abstract: Scale-space behavior of corners is important for developing an efficient corner detection algorithm. In this paper, we analyze the scale-space behavior with the Laplacian of Gaussian (LoG) operator on a planar curve which constructs Laplacian Scale Space (LSS). The analytical expression of a Laplacian Scale-Space map (LSS map) is obtained, demonstrating the Laplacian Scale-Space behavior of the planar curve corners, based on a newly defined unified corner model. With this formula, some Laplacian Scale-Space behavior is summarized. Although LSS demonstrates some similarities to Curvature Scale Space (CSS), there are still some differences. First, no new extreme points are generated in the LSS. Second, the behavior of different cases of a corner model is consistent and simple. This makes it easy to trace the corner in a scale space. At last, the behavior of LSS is verified in an experiment on a digital curve.

Journal ArticleDOI
TL;DR: In this article, an algorithm for the automatic detection of ceiling lightings is developed and tested, the main sections of the algorithm consist of ceiling extraction, point cloud to image conversion, and luminaires detection.

Proceedings ArticleDOI
22 Jun 2015
TL;DR: Tests showed that the proposed pattern recognition method demonstrates no more than 1% recognition error when number of false targets is up to 40 and is suitable for automatic pattern detection in a dense environment of false objects.
Abstract: In this work we present a pattern recognition method based on geometry analysis of a flat pattern. The method provides reliable detection of the pattern in the case when significant perspective deformation is present in the image. The method is based on the fact that collinearity of the lines remains unchanged under perspective transformation. So the recognition feature is the presence of two lines, containing four points each. Eight points form two squares for convenience of applying corner detection algorithms. The method is suitable for automatic pattern detection in a dense environment of false objects. In this work we test the proposed method for statistics of detection and algorithm's performance. For estimation of pattern detection quality we performed image simulation process with random size and spatial frequency of background clutter while both translational (range varied from 200 mm to 1500 mm) and rotational (up to 60°) deformations in given pattern position were added. Simulated measuring system included a camera (4000x4000 sensor with 25 mm lens) and a flat pattern. Tests showed that the proposed method demonstrates no more than 1% recognition error when number of false targets is up to 40.

Proceedings ArticleDOI
01 Oct 2015
TL;DR: A combination of SIFT and Random Sample Consensus (RANSAC) is used to produce panoramic image and the results shows that the projective transformation has a better performance in terms of accuracy.
Abstract: Image registration is a process of determining the geometrical transformation that aligns two or more images taken from different viewpoints and sensors at different times. Scale Invariant Feature Transform (SIFT) method has gained more popularity since it extracts the highest number of features and matching points compared to Speeded-Up Robust Feature (SURF) and Harris Corner Detector at little computational cost. In this paper, a combination of SIFT and Random Sample Consensus (RANSAC) is used to produce panoramic image. In order to reject outliers and estimate the transformation model, affine and projective transformations are used to study the best geometrical transformations methods to be used. The results shows that the projective transformation has a better performance in terms of accuracy.

Proceedings ArticleDOI
01 Jan 2015
TL;DR: The experimental results show that the proposed Harris corner detection algorithm can detect Harris corners of multispectral images efficiently and have significant variations both in spatial domain and spectral domain.
Abstract: The feature analysis plays a more and more important role in the processing of multispectral images, and it is good at acquiring the key information of images. The Harris corner is an important local feature, and it has been generally applied in the processing and analysis of images. However, the existing Harris corner detection algorithms are mainly applied in gray and color images. Therefore, based on the traditional Harris corner detection algorithm for two-dimensional images, this paper develops a Harris corner detection algorithm for three-dimensional multispectral images to acquire the key information of multispectral images. The experimental results show that the proposed Harris corner detection algorithm can detect Harris corners of multispectral images efficiently. These corners are some points that have significant variations both in spatial domain and spectral domain.

Proceedings ArticleDOI
10 Aug 2015
TL;DR: A key point based passive image forensic technique to detect and localize forged regions in a manipulated image and the effectiveness of the proposed method to general geometrical transformations and post processing operations is revealed.
Abstract: Copy-Move Image Forgery is a clean and efficient method to perform digital image forgeries. To localize a forged region, is an important task in detecting such kind of forgeries. This paper proposes a key point based passive image forensic technique to detect and localize forged regions in a manipulated image. Key points are extracted from the image using Harris Corner detector and a specific area around each key point is used for feature extraction using BRISK. The Hamming distance metric is used to detect the distances among the features obtained to facilitate the measurement of similarity. Similarity among feature vectors is detected using Nearest Neighbor Distance Ratio. Finally outliers are removed using RANSAC. The experiment results reveal the effectiveness of the proposed method to general geometrical transformations and post processing operations.

Journal ArticleDOI
TL;DR: A novel approach for image description based on two well-known algorithms: edge detection and blob extraction that provides a mathematical description of each object in the input image and produces a histogram which can be used in various fields of computer vision.
Abstract: In this paper we present a novel approach for image description. The method is based on two well-known algorithms: edge detection and blob extraction. In the edge detection step we use the Canny detector. Our method provides a mathematical description of each object in the input image. On the output of the presented algorithm we obtain a histogram, which can be used in various fields of computer vision. In this paper we applied it in the content-based image retrieval system. The simulations proved the effectiveness of our method.

Patent
15 Apr 2015
TL;DR: A checker corner detection method using a Harris corner detection algorithm to detect a checker image, thus obtaining candidate corners; a coordinate of the candidate corner is very precise and at a sub pixel level; taking each candidate corner as a center so as to obtain a square symmetrical template; using the square symmetric template to process the candidate corners, and removing false corners, thus, obtaining check corners.
Abstract: A checker corner detection method uses a Harris corner detection algorithm to detect a checker image, thus obtaining candidate corners; a coordinate of the candidate corner is very precise and at a sub pixel level; taking each candidate corner as a center so as to obtain a square symmetrical template; using the square symmetrical template to process the candidate corners, and removing false corners, thus obtaining check corners. The checker corner detection method analyzes gray value distributed regularity in checker image corner neighbourhood, uses the square symmetrical template taking the corner as the center, and employs the square symmetrical template to process the candidate corners and removes the false corners so as to obtain the check corners; the method is short in calculation time, and high in detection precision.

Proceedings ArticleDOI
01 Nov 2015
TL;DR: This paper presents a method to localize the corners of cervical vertebrae in a set of 90 lateral cervical radiographs and demonstrates promising results, identifying corners with an average median error of 2.08 mm.
Abstract: The neck (cervical spine) is a flexible part of the human body and is particularly vulnerable to injury. Patients suspected of cervical spine injuries are often imaged using lateral view radiographs. Incorrect diagnosis based on these images may lead to serious long-term consequences. Our overarching goal is to develop a computer-aided detection system to help an emergency room physician correctly diagnose a patient's injury. In this paper, we present a method to localize the corners of cervical vertebrae in a set of 90 lateral cervical radiographs. Haar-like features are computed using intensity and gradient image patches, each of which votes for possible corner position using a modified Hough forest regression technique. Votes are aggregated using two dimensional kernel density estimation, to find the location of the corner. Our method demonstrates promising results, identifying corners with an average median error of 2.08 mm.

Journal ArticleDOI
TL;DR: Wang et al. as discussed by the authors employed the Chebyshev polynomial fitting to estimate the curvature of a planar curve in a continuous way, which achieved promising performance in comparison with three representative corner detectors based on discrete curvature estimation.
Abstract: Extensive contour-based corner detection methods have been proposed to estimate the curvature of a planar curve and most of them estimated it in digital spaces. However, the curvature should be intuitively continuous. Presented is a novel approach which addresses the corner detection issue by employing the Chebyshev polynomial fitting to estimate the curvature in a continuous way. Experimental results demonstrate that the approach achiever a promising performance in comparison with three representative corner detectors based on discrete curvature estimation and two other state-of-the-art methods.

Patent
25 Nov 2015
TL;DR: In this article, a checkerboard corner detection method is combined with a Robust's checkboard detection method, and two groups of checkerboards corners are adopted to serve as filters on the basis of the characteristics of the corners to filter marker images.
Abstract: With an object of solving problems existing in the prior art, the invention provides a fully automatic calibration method for a high performance camera under a complicated background The method is combined with a Robust's checkerboard corner detection method, and two groups of checkerboards corners are adopted to serve as filters on the basis of the characteristics of the corners to filter marker images Eight filters of two types are reduced to four filters of two types, which means that the wave processing amount halves and therefore, the calibration speed increases Further, the method utilizes a Zhang's camera calibration method to mark the camera calibration parameters The method utilizes the marked camera calibration parameters to standardize images to be corrected, and a normalized correlation of the standardized images is calculated, and then sub-pixel precision checkerboard corners are obtained The method then re-projects the coordinate of the checkerboard corners onto image space to obtain a precise coordinate of the corner image The newly obtained coordinate of the corners is then substituted into the Zhang's calibration method to obtain new computed camera parameters By repeating the above steps, a performer can obtain highly precise camera parameters According to the embodiments of the invention, it is possible for a performer to complete automatic corner detection and camera calibration without having to resort to human-machine interactive operations

Journal ArticleDOI
TL;DR: This algorithm is useful for resolving the problem of potential errors due to parallax effects when establishing geometric affine transformation on corners for detecting on buildings with different unknown elevations and has the lower time cost comparing with the other existing algorithms.
Abstract: In order to overcome the difficulty of automatic image registration in image preprocessing, this paper presents an automatic registration algorithm for remote sensing images with different spatial resolutions. The algorithm is studied based on Harris-Laplacian corner detection, which can determine the affine transformation (zoom, rotation, translation) between images of different scales. The corners in the reference and registration images are firstly detected and located by a multi-scale Harris-Laplacian (H-L) corner detector. Secondly, the algorithm chooses SURF (Speeded Up Robust Feature) descriptor to calculate the detected corners descriptors. Then, the multi-resolution corner matching is achieved based on Euclid distance. Finally, according to the LoG (Laplacian Of Gaussian), the scale factor is automatically determined between reference and registration images. A number of remote sensing images are tested, and the experiments show that the studied algorithm can register two remote sensing images of different sizes and resolutions automatically. It also verifies that the algorithm has the lower time cost comparing with the other existing algorithms (e.g. SIFT) within certain detecting accuracy level. This algorithm is also useful for resolving the problem of potential errors due to parallax effects when establishing geometric affine transformation on corners for detecting on buildings with different unknown elevations.

Patent
19 Aug 2015
TL;DR: In this article, the authors proposed an electronic image-stabilizing method based on an infrared thermal imager, which comprises steps of preprocessing, motion estimation, motion filtering, motion compensation and image output.
Abstract: The invention discloses an electronic image-stabilizing method based on an infrared thermal imager. The method comprises steps of preprocessing, motion estimation, motion filtering, motion compensation and image output. In virtue of the above steps, the electronic image-stabilizing method based on the infrared thermal imager saves equipment cost, performs motion estimation by means of corner detection in combination with an optical flow method, smoothens an image motion locus, effectively improves image dithering, and improves an anti-dithering effect.

Journal Article
TL;DR: The result shows that, when the parameters of the Harris corner detection algorithm is improved, it can highlight the corner and background contrast, reduce the complexity of the threshold value to a certain extent, and the result is better than the classical Harris algorithm.
Abstract: Corner is an important image features,we provide theoretical analysis and experimental study of the classic Harris algorithm. We propose a theoretical analysis of parameter values in the improvement,and validate the analysis results based on the related program testing. The result shows that,when the parameters of the Harris corner detection algorithm is improved,it can highlight the corner and background contrast,reduce the complexity of the threshold value to a certain extent,and the result is better than the classical Harris algorithm.

Journal Article
TL;DR: The interpolation algorithm with the use of splines is carried out to recover the length of approximated signal and corner points represent the enormous changes of curvature and is shown the maximum or minimum on the distance curve, therefore, corner points can be detected by calculating the extrimum of distance curve.
Abstract: To achieve successful segmentation of overlapped apples, a segmentation method by using Snake model and corner detectors was presented. As contour is an important basis for detection and recognition of object, and remarkable characteristic of overlapped apples has some typical angular points, which are also called segmentation points and in the target contour. Since Snake model could better converge to target's concave places, Snake model was used to extract overlapped apples' outline. For searching overlapped apples' corner points, corner detection algorithm based distance was proposed: 1) overlapped apples' contour was coded; 2) the distance between contour points and the given ‘center point' was calculated, where ‘center point' was overlapped apples' centroid point for the simplicity of calculation; 3) the distance curve that was get in step 2 is useless as it may engender a lot of spurious corner points. This is caused by small disturbances of small distance, for removing spurious corner points, db1 wavelet was utilized to decomposed original signal at level three, there is a relationship between wavelet transform and digital filter banks. so the wavelet transform can be simply achieved by a tree of digital filter banks. The idea behind filter banks is to divide a signal into two parts: one is the low frequency part and the other is the high frequency part, which could be achieved by a set of filters, the low frequency that is approximate version of the original distance curve in this paper don't contain detail components of original distance and is beneficial to detect true corner points. But the problem with the use of these filters is that each of the two decomposed signals is subjected to downsampling, which simply means throwing away every second data point. After decomposition with three levels, the length of approximated signal reduced, which may cause the miss of the index of original contour point. As for this reason, the approximated signal must be recovered to its original length. In this paper, the interpolation algorithm with the use of splines is carried out to recover the length of approximated signal; 4) corner points represent the enormous changes of curvature and is shown the maximum or minimum on the distance curve, therefore, corner points can be detected by calculating the extrimum of distance curve. Detected corner points need to be selected to determine overlapping positions, a segmentation method based long axis segmentation rule was proposed to choose segmentation line: 1) overlapped apples were divided into uniform two parts approximately by long axis; 2) segmentation line was chose by calculating the distance between bilateral corner points and centroid point. Some split criteria were given as: 1) the direction of the detected points should be opposite, which meant that the detected points from the same region should not be used to split an object; 2) the length of split line should be short. By using these given criteria, the detected corner points were matched to realize the segmentation of overlapped apples. To validate the effectiveness of the algorithm, 20 overlapped apples in nature scenes were tested. Compared with segmentation line obtained by artificial calculation, highest segmentation error of the proposed method is 13.27°, minimum error is 1.20°, and the average error 6.41°. Experimental results show that the proposed segmentation algorithm has a preferable performance, and it is feasible and valid for overlapped apple segmentation in nature scenes.

Journal ArticleDOI
TL;DR: Results showed the proposed method by iterative procedure can make the impact of slight distortion around image center negligible and the average of distortion residual of one line is almost 0.3 pixels.
Abstract: This paper proposes a detection method of chessboard corner to correct camera distortions -including radial distortion, decentering distortion and prism distor- tion. This proposed method could achieve high corner detection rate. Then we used iterative procedure to optimize distortion parameter to minimize distortion residual. In this method, first, non-distortion points are evaluated by four points near image center; secondly, Levenberg-Marquardt nonlinear optimization algorithm was adopted to calculate distortion parameters, and then to correct image by these parameters; thirdly, we calculated corner points on the corrected image, and repeated previous two steps until distortion parameters converge. Results showed the proposed method by iterative proce- dure can make the impact of slight distortion around image center negligible and the average of distortion residual of one line is almost 0.3 pixels.

Book ChapterDOI
01 Jan 2015
TL;DR: The idea of using the super-resolution algorithms for the self-localization and vision based navigation of autonomous mobile robots is discussed and some simplified systems such navigation is based on the edge and corner detection and binary image analysis, which could be troublesome for low resolution images.
Abstract: In the paper the idea of using the super-resolution algorithms for the self-localization and vision based navigation of autonomous mobile robots is discussed. Since such task is often limited both by the limited resolution of the mounted video camera as well as the available computational resources, a typical approach for video based navigation of mobile robots, similarly as many small flying robots (drones), is using low resolution cameras equipped with average class lenses. The images captured by such video system should be further processed in order to extract the data useful for real-time control of robot’s motion. In some simplified systems such navigation, especially in the within an enclosed environment (interior), is based on the edge and corner detection and binary image analysis, which could be troublesome for low resolution images.

Book ChapterDOI
01 Jan 2015
TL;DR: Object detection methods that utilize RGB cameras are used to accurately identify objects in the real world, but they do not consider shape and three-dimensional characteristics of the object.
Abstract: Object detection is a challenging field of research in computer vision. Research approaches have become increasingly popular in overcoming the challenges of object detection like occlusions, changes in scale, rotation, and illumination. Object detection methods that utilize RGB cameras are used to accurately identify objects in the real world, but they do not consider shape and three-dimensional characteristics of the object. Recognizing the objects in 3D is not an easy task for computers, like as in humans. Robust features like shape, color, size, etc., are necessary for 3D object detection for ensuring accuracy.

Journal ArticleDOI
30 Mar 2015-Minerals
TL;DR: The experimental results show that the new method can produce good bubble delineation results automatically and can alleviate the over-segmentation problem effectively and its processing speed can also meet the online measurement requirements.
Abstract: Froth image segmentation is an important and basic part in an online froth monitoring system in mineral processing. The fast and accurate bubble delineation in a froth image is significant for the subsequent froth surface characterization. This paper proposes a froth image segmentation method combining image classification and image segmentation. In the method, an improved Harris corner detection algorithm is applied to classify froth images first. Then, for each class, the images are segmented by automatically choosing the corresponding parameters for identifying bubble edge points through extracting the local gray value minima. Finally, on the basis of the edge points, the bubbles are delineated by using a number of post-processing functions. Compared with the widely used Watershed algorithm and others for a number of lead zinc froth images in a flotation plant, the new method (algorithm) can alleviate the over-segmentation problem effectively. The experimental results show that the new method can produce good bubble delineation results automatically. In addition, its processing speed can also meet the online measurement requirements.