Curvature Scale Space Research Articles

Curvature Scale Space LiDAR Odometry And Mapping (LOAM)

The LiDAR Odometry and Mapping (LOAM) algorithm ranks in second place in the Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI), Visual Odometry/SLAM Evaluations. It utilizes a feature extraction algorithm based on the evaluation of the curvature of points under test, to produce estimated smooth and non-smooth regions within typically laser based Point Cloud Data (PCD). This feature extractor (FE) however, does not take into account PCD spatial or detection uncertainty, which can result in the divergence of the LOAM algorithm. Therefore, this article proposes the use of the Curvature Scale Space (CSS) algorithm as a replacement for LOAM’s current feature extractor. It justifies the substitution, based on the CSS algorithm’s similar computational complexity but improved feature detection repeatability. LOAM’s current feature extractor and the proposed CSS feature extractor are tested and compared with simulated and real data, including the KITTI odometry-laser data set. Additionally, a recent deep learning based LiDAR Odometry (LO) algorithm, the Convolutional Auto-Encoder (CAE)-LO algorithm, will also be compared, using this data set, in terms of its computational speed and performance. Performance comparisons are made based on the Absolute Trajectory Error (ATE) and Cardinalized Optimal Linear Assignment (COLA) metrics. Based on these metrics, the comparisons show significant improvements of the LOAM algorithm with the CSS feature extractor compared with the benchmark versions.

AN IMPROVED IMAGE REGISTRATION ALGORITHM FOR THERMAL INFRARED AND PANCHROMATIC IMAGE BASED ON GEOMETRIC STRUCTURAL PROPERTIES

Abstract. In order to improve the image registration accuracy of heterogeneous remote sensing images with large radiation differences, an improved image registration algorithm for thermal infrared and panchromatic images is proposed. This method uses the phase consistency of intensity and direction to construct a geometric structure feature descriptor called the Histogram of Oriented Phase Congruency (HOPC) for image registration. It employs the Curvature Scale Space (CSS) corner detection to concentrate and extract feature points in contour areas, and defines a similarity metric (called HOPCn) based on the Euclidean distance between the descriptors for high-precision heterogeneous images registration. Five sets of panchromatic and thermal infrared images were selected for verification. The results show that the improved algorithm can match the homonymy points around the image contour, and the Correct Matching Rate (CMR) is improved by 2.7% and the Root Mean Square Error (RMSE) is improved by 3.7% compared to the original algorithm. This proves that the proposed method has good robustness under conditions of large radiation differences.

A robust visible and infrared image matching algorithm for power equipment based on phase congruency and scale-invariant feature

Dense and accurate matching for visible and infrared images of power equipment is crucial to intelligent diagnosis system of power grid, but existing matching methods usually fail in aligning visible and infrared image pairs because of significant intensity, resolution and viewpoint differences. In this paper, we propose a matching algorithm based on phase congruency and scale-invariant feature to address this problem. The proposed method consists of four steps. First, the maximum moment map of phase congruency of input image is computed based on phase congruency theory, which is then used to enhance the raw image. Second, Canny operator and contour tracking method are employed to detect image contours and scale-invariant feature points are extracted by the curvature scale space (CSS) corner detector. Third, the novel histogram of phase congruency orientation (HPCO) descriptors based on phase congruency information are computed for all feature points. Finally, a set of preliminary matches is obtained by the bidirectional matching, and refinement procedures are implemented to achieve dense and accurate matching results. We conduct the experiments on public available dataset. Experimental results show that the proposed method can robustly match feature points in visible and infrared image pairs of power equipment in spite of intensity, resolution and viewpoint differences, and achieve favorable performance compared to state-of-the-art approaches.

Confusion Matrix as Performance Measure for Corner Detectors

Nowadays, corner detection algorithms have been proposed by several researchers who described them contrarily, depending on their respective viewpoints to obtain the data and information as a human eye does. Basically, no researchers have come up with a technique to compare corner detectors with another’s. Thus, this study proposed to adapt the confusion matrix technique as a performance measure for corner detectors. The judgement accuracy of every corner detector will only be pleased if the actual corner points are already known. Therefore, this study is attracted to explore the accuracy of corner detectors, namely the Global and Local Curvature Scale space (GLCSS), Affine Resilient Curvature Scale Space (ARCSS), and Harris. These corner detectors were analysed using the nine characters selected from Jawi, Chinese, and Tamil characters, three characters each, respectively. This study specifically detected the true corners for these characters using the determined corner detectors. The actual corner of all these characters was confirmed through a survey of twenty respondents. The majority of marked corners by respondents were considered actual corner points. Then, the input image for all characters was converted into a grayscale image. Every image will undergo pre-processing step, the process of boundary extraction using Canny edge detector. Thus, the edge image was extracted to get the corner point by applying the corner detectors, and the corner point detected was marked on that image. Above and beyond, the study aims to introduce a confusion matrix approach as a performance measure to carry out the most outstanding algorithm in detecting the true corner points for all the tested characters. From the evaluation, GLCSS and Harris algorithms have shown good accuracy. Henceforth, the study is not trying to judge the goodness of each corner detector but only to introduce confusion matrix as a tool that can be considered to measure the performance of the corner detector.

A new registration method with improved phase congruency for application to GPR images

ABSTRACT In this letter, we propose an edge detection algorithm and a framework of points matching for ground penetrating radar (GPR) images to achieve the B-Scan image registration of GPR. In our proposed algorithm, the robust edge feature information of GPR images is detected by the improved phase congruency model. Then, the feature points are detected by the GPR image’s edge feature in the curvature scale space. Finally, a novel similarity metric is defined for feature point registration. Experimental results show that compared to the classic registration method and the state-of-the-art feature points registration methods, the proposed method shows the best matching performance and robustness in GPR images.

Computer-Assisted Image Processing System for Early Assessment of Lung Nodule Malignancy.

Simple SummaryLung cancer is the second most common cancer in men after prostate cancer and in women after breast cancer, but it is the leading cause of cancer death among both genders. This manuscript proposes a new computer-aided diagnosis system that uses only a single computed tomography scan to diagnose the pulmonary nodule as benign or malignant. This system helps in the early detection of the pulmonary nodules and shows its ability to identify the pulmonary nodules precisely.Lung cancer is one of the most dreadful cancers, and its detection in the early stage is very important and challenging. This manuscript proposes a new computer-aided diagnosis system for lung cancer diagnosis from chest computed tomography scans. The proposed system extracts two different kinds of features, namely, appearance features and shape features. For the appearance features, a Histogram of oriented gradients, a Multi-view analytical Local Binary Pattern, and a Markov Gibbs Random Field are developed to give a good description of the lung nodule texture, which is one of the main distinguishing characteristics between benign and malignant nodules. For the shape features, Multi-view Peripheral Sum Curvature Scale Space, Spherical Harmonics Expansion, and a group of some fundamental morphological features are implemented to describe the outer contour complexity of the nodules, which is main factor in lung nodule diagnosis. Each feature is fed into a stacked auto-encoder followed by a soft-max classifier to generate the initial malignancy probability. Finally, all these probabilities are combined together and fed to the last network to give the final diagnosis. The system is validated using 727 nodules which are subset from the Lung Image Database Consortium (LIDC) dataset. The system shows very high performance measures and achieves , , and for the accuracy, sensitivity, and specificity, respectively. This high performance shows the ability of the system to distinguish between the malignant and benign nodules precisely.

Cloning detection scheme based on linear and curvature scale space with new false positive removal filters

Cloning detection scheme based on linear and curvature scale space with new false positive removal filters

Registration Method for Power Equipment Infrared and Visible Images Based on Improved Curvature Scale Space Algorithm

Registration Method for Power Equipment Infrared and Visible Images Based on Improved Curvature Scale Space Algorithm

Box filtering for real-time curvature scale-space computation

Curvature scale-space (CSS) analysis is an important technique for contour-based object recognition in digital images. To compute the CSS for a given contour, it is systematically convolved (smoothed) with Gaussians with increasing standard deviation. The convolutions are computationally expensive, especially for large and high resolution contours, but can be approximated using box filtering (also known as mean and average filtering). Together with running sums, the convolutions can be accelerated by 2–3 magnitudes without significant loss of precision. Nonetheless, box filtering has not been systematically investigated in connection with CSS computation. In this work, we present a theoretical and experimental analysis of different box-filtering techniques in this context and conclude which is the most efficient implementation. Based on this, the CSS of a contour can be computed in real time with high precision.

SCN: A Novel Shape Classification Algorithm Based on Convolutional Neural Network

Shape classification and matching is an important branch of computer vision. It is widely used in image retrieval and target tracking. Shape context method, curvature scale space (CSS) operator and its improvement have been the main algorithms of shape matching and classification. The shape classification network (SCN) algorithm is proposed inspired by LeNet5 basic network structure. Then, the network structure of SCN is introduced and analyzed in detail, and the specific parameters of the network structure are explained. In the experimental part, SCN is used to perform classification tasks on three shape datasets, and the advantages and limitations of our algorithm are analyzed in detail according to the experimental results. SCN performs better than many traditional shape classification algorithms. Accordingly, a practical example is given to show that SCN can save computing resources.

Research on Contour Feature Extraction Algorithms for Arc-shaped Groove of Safety Seat Base

The spatial position of the seat base is complicated. The contour structure of arc groove is complex and its contour feature extraction is difficult. Aiming at the above problems, this paper presents an improved corner detection algorithm based on Curvature Scale Space (CSS), which uses adaptive local curvature and dynamic angle threshold to accurately extract corner points of arc groove contour. The correct corner positioning error of the algorithm proposed in this paper is only 0.99 pixel. The experimental results show that this method can effectively implement contour feature extraction.

Curvature scale space-based signatures for electrical load classification in NILM

Non-intrusive load monitoring (NILM) systems have gained extensive interest due to their potential role regarding power savings for residential customers. These systems, which are mostly based on stages of detection and classification of transients on aggregated signals, rely heavily on load signatures. In the literature, the image-based voltage–current (V–I) trajectory representation is claimed as the most effective individual steady-state signature for appliance classification. However, this representation inherits some drawbacks from its generation process. This work then proposes two steady-state appliance signatures derived from the curvature scale space of V–I trajectories. These signatures aim to improve the image representations of V–I trajectories by encompassing structural elements related to the general shape of such trajectories as well as some characteristics neglected during their generation. A group of load signatures formed from the proposed signatures was evaluated on direct load classification and load disaggregation scenarios for publicly available datasets. The achieved results surpassed the sole employment of a reference image-based V–I signature for all the test scenarios executed. Also, some of the evaluated signatures outperformed all known proposals that are exclusively based on steady-state signatures for load classification on a given benchmarking dataset.

Automatic segmentation algorithm for breast cell image based on multi-scale CNN and CSS corner detection

As the shapes of breast cell are diverse and there is adherent between cells, fast and accurate segmentation for breast cell remains a challenging task. In this paper, an automatic segmentation algorithm for breast cell image is proposed, which focuses on the segmentation of adherent cells. First of all, breast cell image enhancement is carried out by the staining regularization. Then, the cells and background are separated by Multi-scale Convolutional Neural Network (CNN) to obtain the initial segmentation results. Finally, the Curvature Scale Space (CSS) corner detection is used to segment adherent cells. Experimental results show that the proposed algorithm can achieve 93.01% accuracy, 93.93% sensitivity and 95.69% specificity. Compared with other segmentation algorithms of breast cell, the proposed algorithm can not only solve the difficulty of segmenting adherent cells, but also improve the segmentation accuracy of adherent cells.

A Contour Angle Orientation for Power Equipment Infrared and Visible Image Registration

Automatic registration for infrared, and visible images of power equipment has become a challenging work in intelligent diagnosis system of the power grid. Existing registration methods usually fail in accurately aligning power equipment infrared, and visible images because of resolutions, spectrums, and viewpoints differences. To solve this problem, we propose a novel main orientation of feature points named contour angle orientation (CAO), and describe an automatic infrared, and visible image registration method named CAO-Coarse to Fine (CAO-C2F). CAO is based on the contour feature of images, and invariant to images viewpoints, and scales differences. C2F is a feature matching method to obtain correct matches. Our proposed CAO-C2F method includes four steps. First, feature points in contours are extracted by the curvature scale space (CSS) corner detector based on local, and global curvature. Second, the CAO of each feature point is computed as the main orientation. Third, modified scale-invariant feature transform (SIFT) descriptors on the main orientations are extracted, and matched by bilateral matching. Finally, accurate matches are obtained by applying the C2F method. Registration experiments on a self-established images dataset show our proposed CAO-C2F method accurately aligns images, and outperforms other state-of-arts in terms of precision, recall, and root-mean-square error.

Open Curvature Scale Space Matching for Coronary Artery Identification in X-Ray Angiographic Images

Identification of the coronary artery in sequence angiograms is an important issue in the analysis and treatment of coronary angiography data, due to the ambiguities in the coronary vessels resulting from cardiac motion and the differences in the projection angles in angiograms. In this paper, we propose a novel coronary artery identification method based on curvature scale space image matching. First, points on the centerline of each coronary artery branch are convolved with Gaussian kernel functions of different scales to obtain the corresponding curvature scale space (CSS) image during the curve evolution. Then, a CSS image matching method for open curves, which uses a combination of open feature vectors and closed feature vectors, is proposed. Finally, curve matching transitivity is used to solve the problem of the inability to recognize the coronary artery when the difference in the imaging angles is too large. The experimental results demonstrate that the proposed method can recognize every coronary artery in different views, even when the difference in the imaging angles is 29.8°.

Method for segmentation of overlapping fish images in aquaculture

Individual fish segmentation is a prerequisite for feature extraction and object identification in any machine vision system. In this paper, a method for segmentation of overlapping fish images in aquaculture was proposed. First, the shape factor was used to determine whether an overlap exists in the picture. Then, the corner points were extracted using the curvature scale space algorithm, and the skeleton obtained by the improved Zhang-Suen thinning algorithm. Finally, intersecting points were obtained, and the overlapped region was segmented. The results show that the average error rate and average segmentation efficiency of this method was 10% and 90%, respectively. Compared with the traditional watershed method, the separation point is accurate, and the segmentation accuracy is high. Thus, the proposed method achieves better performance in segmentation accuracy and effectiveness. This method can be applied to multi-target segmentation and fish behavior analysis systems, and it can effectively improve recognition precision. Keywords: aquaculture, image processing, overlapping segmentation, corner detection, improved Zhang-Suen algorithm DOI: 10.25165/j.ijabe.20191206.3217 Citation: Zhou C, Lin K, Xu D M, Liu J T, Zhang S, Sun C H, et al. Method for segmentation of overlapping fish images in aquaculture. Int J Agric & Biol Eng, 2019; 12(6): 135–142.

Corner detection based on gradient ghost imaging

Gradient ghost imaging (GGI) is a new imaging method that can directly extract the edge of a target from the correlation of light intensity fluctuations. However, there are problems of poor image quality and practicability in GGI. The corner point is an important target feature and has important applications in the fields of image processing and machine vision. In this paper, we propose a corner detection scheme that combines a corner detection algorithm based on curvature scale space with a GGI system, which can directly extract corner points from the reconstruction results of ghost imaging. A simulation and experimental results show that our method can acquire precise and robust corner information of an unknown target in undersampled cases, which promotes the practical development of ghost imaging.

Computer Algorithm for Archaeological Projectile Points Automatic Classification

The manual archaeological projectile point morphological classification is an extensive and complex process since it involves a large number of categories. This article presents an algorithm that automatically makes this process, based on the projectile point digital image and using a classification scheme according to global archaeological approaches. The algorithm supports different conditions such as changes in scale and quality of the image. Moreover, it requires only a uniform background and an approximate north--south projectile point orientation. The principal computer methods that compose the algorithm are the curvature scale space map (CSS-map), the gradient contour on the projectile point, and the support vector machines (SVM) algorithm. Finally, the classifier was trained and tested on a dataset of approximately 800 projectile points images, and the results have shown a better performance than other shape descriptors such as Pyramid of Histograms of Orientation Gradients (PHOG), Histogram of Orientation Shape Context (HOOSC) (both used in a bag-of-words context), and geometric moment invariants (Hu moments).

Determination of Building Model Key Points Using Multidirectional Shaded Relief Images Generated from Airborne LiDAR Data

Light detection and ranging (LiDAR) data collected from airborne laser scanner system is one of the major sources to reconstruct Earth’s surface features. This paper presents a method for detecting model key points (MKPs) of the buildings using LiDAR point clouds. The proposed approach utilizes shaded relief images (SRIs) derived from the LiDAR data. The SRIs based on the concept of the shape from shading could provide unique information about individual surface patches of the building roofs. The main advantage of the proposed approach is to detect directly MKPs, which are primitives for 3D building modeling, without segmenting point clouds. Depending on the location of the light source, the SRIs are created differently. Therefore, integration of the multidirectional SRIs created from different locations of the light source could provide more reliable results. In addition, the vertical exaggeration (i.e., scaling Z-coordinates) is also beneficial because constituent surface patches of the roofs in the SRIs created with vertically exaggerated LiDAR data are more distinguishable. To determine the MKPs of the roofs, building data was separated from other objects using modified marker-controlled watershed algorithm in accordance with criteria to specify buildings such as area, height, and standard deviation. This process could remove the unnecessary objects such as trees, vegetation, and cars. The curvature scale space (CSS) corner detector was used to determine MKP since this method is robust to geometric changes such as rotation, translation, and scale. The proposed method was applied to simulated and real LiDAR datasets with various roof types. The experimental results show that the proposed method is effective in determining MKPs of various roof types with high level of detail (LoD).

An almost complete curvature scale space representation: Euclidean case

Here, we intend to propose local shape curve features which are invariant under planar Euclidean transformations and independent with respect to the original curve parameterization. The present work generalizes the family of Curvature Scale Space descriptors in order to increase the shape information quantity to tend to the completeness property. For this, a more pragmatic criterion is introduced in this paper which we call the almost completeness. We define it as a pre-completeness for a given resolution of features. Such descriptors are formed by the curvatures on the set of curve points obtained from the antecedents of different curvature levels. This level set is fixed with a given rule. The idea of the almost completeness is to make a compromise between the cardinal of the set of curvature’s levels and the optimal number of scales. The rule is submitted to an unsupervised statistical study and the scales are obtained with a spectral analysis. Experiments are conducted on several known datasets. Promising results in the sense of shape retrieval and shape recognition rates are demonstrated.