Feature Descriptor Matching Research Articles

Simultaneous localization and mapping of mobile robots with multi-sensor fusion

Abstract Simultaneous localization and map building of mobile robots is an important research direction in robotics, and multi-sensor fusion is one of the key technologies for mobile robots to achieve autonomous navigation. In this paper, firstly, based on the framework of the SLAM system, we analyze the extraction of FAST feature points and BRIEF feature descriptor matching and propose the CSD-ORB algorithm. Secondly, the FKCS-LD algorithm based on the fusion map key frame and center selection strategy is proposed. The suitable multi-sensor fusion SLAM system is selected to build the back-end optimization algorithm based on the bit-pose map. Finally, we complete the comparison experiments of the multi-sensor fusion SLAM system on standard data sets, robot hardware assembly and software environment construction, and actual field testing. The experimental data show that the trajectory calculation accuracy can be further improved by using local map tracking, i.e., the current frame and local keyframe to calculate the bit pose, in which the maximum error is 6 cm, the root mean square error is reduced to 3.6 cm, the maximum value of local optimization is 0.66034, the minimum value is 0.146232, the average is 0.32812, and the median is 0.33023. The experiment proves that the multi-sensor fusion SLAM system can achieve centimeter-level accuracy and has global consistency and real-time performance, and can effectively complete the task of simultaneous robot positioning and map building.

Research on Image Mosaic Method Based on Fracture Edge Contour of Bone Tag

Damaged edges of bone tag images contain external factors such as impurities and damage, which affect the stitching process and lead to repair errors. Therefore, this paper proposes a stitching method based on image edge position feature matching. The objective is to improve the accuracy of image stitching by matching feature points based on the position of the image edge pixel so as to solve the accurate stitching of broken edge contour. In the first step of this method, the image containing the broken edge is preprocessed by edge detection, and the location of the broken edge pixel is proposed. Secondly, the feature descriptors were calculated to extract the shape and texture information of the feature points on the fracture edge. Finally, the feature points are optimized by minimum correction and image mosaic is carried out. In terms of image stitching, pre-registration is performed by finding the feature descriptors that are most similar to the edge of the optimum fracture surface profile. The matching operator is added to the overlapping region to obtain the corrected image, and the panoramic image mosaic of the image fracture surface is performed. The experimental results show that feature descriptor matching can ensure the integrity of the fracture, improve the matching accuracy, optimize the uneven deformation of the fracture, ensure the quality of image stitching, and reduce the degree of image distortion.

Salient keypoint-based copy–move image forgery detection

ABSTRACT Copy–move forgery is one of the most common image forgeries in digital images. In copy–move forgery, some object or region of an object is copied and duplicated in other parts of the same image. Hence, there is a need to develop an accurate and robust forgery detection approach for various image forensics applications. In this article, an improved salient keypoint selection approach for copy–move forgery detection has been proposed. Scale-Invariant Feature Transform (SIFT) and KAZE keypoint features have been extracted from the input image and salient keypoints have been selected for improving the robustness of the proposed algorithm. Salient keypoint selection reduces the feature descriptor matching time for finding region duplication in the given image. To improve the detection accuracy of the proposed approach, selective search-based region proposals have been introduced to create a bounding box on the input image. Feature descriptor matching is performed between keypoints which are located inside two different bounding boxes. The proposed approach has been evaluated on two benchmark datasets, CoMoFoD and MICC-F220 and detection results outperformed state-of-the-art techniques under different geometric transformations and post-processing operations.

3DHoPD: A Fast Low-Dimensional 3-D Descriptor

Three-dimensional feature descriptors are heavily employed in various 3-D perception applications to find keypoint correspondences between two point clouds. The availability of mobile devices equipped with depth sensors compels the developed applications to be both memory and computationally efficient. Toward this, in this letter, we present 3DHoPD, a new low-dimensional 3-D feature descriptor that is extremely fast to compute. The novelty lies in compactly encoding the “3-D” keypoint position by transforming it to a new 3-D space, where the keypoints arising from similar 3-D surface patches lie close to each other. Then, we propose histograms of point distributions (HoPD) to capture the neighborhood structure, thus forming 3DHoPD (3D+HoPD). We propose a tailored feature descriptor matching technique, wherein the “3-D” keypoint position in the new 3-D space is used to remove false positive matches, effectively reducing the search space by 90%, and then, the exact match is found using the “HoPD” descriptor. Experimental evaluation on multiple publicly available datasets shows that 3DHoPD is robust to noise and offers stable and competitive keypoint matching performance to the existing state-of-the-art 3-D descriptors with similar dimensionality across datasets, while requiring dramatically low-computational time (10 $\times$ faster). The source code and additional experimental results are available at https://sites.google.com/site/3dhopd/

B-SHOT: a binary 3D feature descriptor for fast Keypoint matching on 3D point clouds

We present the first attempt in creating a binary 3D feature descriptor for fast and efficient keypoint matching on 3D point clouds. Specifically, we propose a binarization technique and apply it on the state-of-the-art 3D feature descriptor, SHOT (Salti et al., Comput Vision Image Underst 125:251–264, 2014) to create the first binary 3D feature descriptor, which we call B-SHOT. B-SHOT requires 32 times lesser memory for its representation while being six times faster in feature descriptor matching, when compared to the SHOT feature descriptor. Next, we propose a robust evaluation metric, specifically for 3D feature descriptors. A comprehensive evaluation on standard benchmarks reveals that B-SHOT offers comparable keypoint matching performance to that of the state-of-the-art real valued 3D feature descriptors, albeit at dramatically lower computational and memory costs.

Fast Underwater Image Mosaicing through Submapping

One of the most important features of mobile robots is their capability to gather data from areas beyond human reach. This capability has increased the demand for the use of robots undertaking exploration tasks, which has naturally led to the need for efficient methods to process the obtained data. Image mosaicing is a useful tool for obtaining a high-resolution visual representation of a large area that has been explored using optical sensors. In this paper, we present an efficient image mosaicing approach that utilizes submapping methods to obtain a map of a surveyed area with reduced computational effort. The approach uses a modified agglomerative hierarchical clustering method to form submaps according to similarity information obtained through feature descriptor matching, and takes advantage of this submapping to reduce the computation and time costs. Comparative results on real challenging underwater datasets are presented.

Feature-based visual tracking for agricultural implements

Abstract: Systems which utilize implement-mounted cameras for machinery feedback, such as row crop cultivators and sectional sprayers, can be upgraded to provide high-accuracy ground speed and tracking data using visual tracking algorithms. Vector data produced by visual tracking can be incorporated into control systems to compensate for implement dynamics in complement with RTK-GNSS receivers and other sensors. Variations of the SURF, SIFT, and ORB feature-descriptor algorithms were evaluated using a dataset of 640×480 pixel videos on six surfaces (gravel, asphalt, grass, seedlings, residue, and pasture) for speeds from 1 to 5 m/s. Feature-descriptor matching of consecutive video frames was tested using two methods of the k-Nearest Neighbors (kNN) algorithm: (1) 1NN with cross-checking, and (2) 2NN with the ratio-test. Ground speed and tracking direction were calculated using a fast histogram filter to reject outliers between frames. Compared to RTK-GNSS, ORB with CLAHE pre-processing (CLORB) and 1NN cross-checking was found to be the most robust with respect to real-time applications. For 95% of measurements, CLORB achieved an error of 0.23 m/s. Similar accuracy was achieved with SURF, U-SURF, and SIFT, but CLORB was capable of producing vector data in real-time (approximately 25 Hz), whereas SURF, U-SURF, and SIFT were only capable of 15 Hz or less.

Hierarchical ensemble of background models for PTZ-based video surveillance.

In this paper, we study a novel hierarchical background model for intelligent video surveillance with the pan-tilt-zoom (PTZ) camera, and give rise to an integrated system consisting of three key components: background modeling, observed frame registration, and object tracking. First, we build the hierarchical background model by separating the full range of continuous focal lengths of a PTZ camera into several discrete levels and then partitioning the wide scene at each level into many partial fixed scenes. In this way, the wide scenes captured by a PTZ camera through rotation and zoom are represented by a hierarchical collection of partial fixed scenes. A new robust feature is presented for background modeling of each partial scene. Second, we locate the partial scenes corresponding to the observed frame in the hierarchical background model. Frame registration is then achieved by feature descriptor matching via fast approximate nearest neighbor search. Afterwards, foreground objects can be detected using background subtraction. Last, we configure the hierarchical background model into a framework to facilitate existing object tracking algorithms under the PTZ camera. Foreground extraction is used to assist tracking an object of interest. The tracking outputs are fed back to the PTZ controller for adjusting the camera properly so as to maintain the tracked object in the image plane. We apply our system on several challenging scenarios and achieve promising results.

Real-Time Feature Descriptor Matching via a Multi-Resolution Exhaustive Search Method

Feature descriptor matching plays an important role in many computer vision applications. This paper presents a novel fast linear exhaustive search algorithm combined with a multi-resolution candidate elimination technique to deal with this problem efficiently. The proposed algorithm is inspired from the existing multi-resolution image retrieval approaches, but releasing the requirement on a norm-sorted database with pre-computed multi-resolution tables. This helps to increase the applicability of the proposed method. Moreover, the computations of candidate elimination are fully performed using a simple L 1 distance metric, which is able to speedup the entire search process without loss of accuracy. This property leads to an accurate feature descriptor matching algorithm with real-time performance, which will be validated in the experiments by testing with the matching of SURF descriptors.

Stereo vision based SLAM using Rao-Blackwellised particle filter

We present an algorithm which can realize 3D stereo vision simultaneous localization and mapping (SLAM) for mobile robot in unknown outdoor environments, which means the 6-DOF motion and a sparse but persistent map of natural landmarks be constructed online only with a stereo camera. In mobile robotics research, we extend FastSLAM 2.0 like stereo vision SLAM with "pure vision" domain to outdoor environments. Unlike popular stochastic motion model used in conventional monocular vision SLAM, we utilize the ideas of structure from motion (SFM) for initial motion estimation, which is more suitable for the robot moving in-large-scale outdoor, and textured environments. SIFT features are used as natural landmarks, and its 3D positions are constructed directly through triangulation. Considering the computational complexity and memory consumption, Bkd-tree and Best-Bin-First (BBF) search strategy are utilized for SIFT feature descriptor matching. Results show high accuracy of our algorithm, even in the circumstance of large translation and large rotation movements.