Search

Deep neural networks have achieved great progress in 3D scene understanding. However, recent methods mainly focused on objects with canonical orientations in contrast with random postures in reality. In this letter, we propose a hierarchical neural network, named Local Frame Network (LFNet), based on the local rotation invariant coordinate frame for robust point cloud analysis. The local point patches in different orientated objects are transformed into an identical distribution based on this coordinate frame, and the transformed coordinates are taken as input features to eliminate the influence of rotations at the input level . Meanwhile, a discrete convolution operator is defined in the constructed coordinate frame to extract rotation invariant features from local patches, which can further remove the influence of rotations at the convolution level . Moreover, a Spatial Feature Encoder (SFE) module is utilized to perceive the spatial structure of the local region. Mathematical analysis and experimental results on two public datasets demonstrate that the proposed method can eliminate the influence of rotations without data augmentation and outperforms other state-of-the-art methods.

Loop closure detection is an important component of visual simultaneous localization and mapping (SLAM). However, most existing loop closure detection methods are vulnerable to complex environments and use limited information from images. As higher-level image information and multi-information fusion can improve the robustness of place recognition, a semantic–visual–geometric information-based loop closure detection algorithm (SVG-Loop) is proposed in this paper. In detail, to reduce the interference of dynamic features, a semantic bag-of-words model was firstly constructed by connecting visual features with semantic labels. Secondly, in order to improve detection robustness in different scenes, a semantic landmark vector model was designed by encoding the geometric relationship of the semantic graph. Finally, semantic, visual, and geometric information was integrated by fuse calculation of the two modules. Compared with art-of-the-state methods, experiments on the TUM RBG-D dataset, KITTI odometry dataset, and practical environment show that SVG-Loop has advantages in complex environments with varying light, changeable weather, and dynamic interference.

Existing 3D single object trackers (SOTs) of a point cloud all apply downscaling when extracting features from points. This operation leads to a loss of spatial and structural information, degrading tracking performance of sparsely distributed and small-scale objects. To address this problem, a structure aware SOT of a point cloud is proposed. Specifically, the backbone network is combined with the auxiliary network to learn point-wise representations. During the training stage, the subsidiary network is used to perform additional tasks and supervisions, which guides the backbone network to extract discriminative structural features. During the inference stage, this network part is detached to meet a real-time requirement as well as to ensure the tracking accuracy. In addition, the impacts of the quantity setting of the input point cloud and re-initiation strategy are discussed; these are significant to the performance but have been ignored by former works. The experimental results show that the proposed method has a distinct improvement even if the tracked object is sparse and small scale.

3D object detection is a challenging problem in 3D computer vision. In this paper, a Point-Voxel Feature Encoder (PVFE) is proposed to obtain the spatial context information between points and voxels for 3D object detection in point clouds. Firstly, the point cloud is converted into 3D voxel grids to avoid loss of spatial structure information in projecting. Then, voxel-wise features are learned through PVFE, which is consisting of stacked fully connected layer. Last, voxel-wise features are processed by the Sparse Convolutional Layers and as the input of RPN which performs category prediction and bounding box regression. Experiments of car, bicyclist and pedestrian detection have been conducted respectively on the KITTI benchmark. Experimental results have clearly shown the superiority of the proposed network.

Golay complementary waveforms are long suffered from dissatisfactory Doppler resolution in target detection. In this paper, we develop a signal processing processor which comparing and pointwise thresholding the radar echo wave of two separate pairs of Golay complementary waveforms. The effectiveness of this pointwise thresholding processor (PTP) in terms of better Doppler resolution than conventional design of Golay complementary waveforms is proved by simulating and analyzing the output Delay-Doppler maps.

Traditional auditory features merely present the amplitude characteristics of target signals in frequency domain. Such features are susceptible to environmental noise, resulting in significant degradation of recognition stability. Inspired by instantaneous information applied in speech signal processing field, this paper proposed a feature extraction method using sub-based instantaneous frequency. A fast instantaneous frequency information extraction algorithm is proposed with the normalized Gammatone filterbanks. Experiments confirm that the proposed feature extraction method can effectively maintain the recognition accuracy under low SNR conditions while reduce the computation cost.

Three-dimensional (3D) object detection is an important research in 3D computer vision with significant applications in many fields, such as automatic driving, robotics, and human&ndash, computer interaction. However, the low precision is an urgent problem in the field of 3D object detection. To solve it, we present a framework for 3D object detection in point cloud. To be specific, a designed Backbone Network is used to make fusion of low-level features and high-level features, which makes full use of various information advantages. Moreover, the two-dimensional (2D) Generalized Intersection over Union is extended to 3D use as part of the loss function in our framework. Empirical experiments of Car, Cyclist, and Pedestrian detection have been conducted respectively on the KITTI benchmark. Experimental results with average precision (AP) have shown the effectiveness of the proposed network.

In recent years, deep learning has become a popular method for 3D object recognition with point clouds. In this work, we introduce a multi-resolution feature fusion convolution neural network using point cloud data for 3D object recognition. Experiments are conducted on ModelNet40 dataset. It achieves better accuracy with 86.9% for 3D object recognition on point cloud data through four different feature fusion. Experimental results have demonstrated the superior performance of the proposed multi-resolution feature fusion network.

Semantic segmentation of 3D scenes is a fundamental problem in 3D computer vision. In this paper, we propose a deep neural network for 3D semantic segmentation of raw point clouds. A multi-scale feature learning block is first introduced to obtain informative contextual features in 3D point clouds. A global and local feature aggregation block is then extended to improve the feature learning ability of the network. Based on these strategies, a powerful architecture named 3DMAX-Net is finally provided for semantic segmentation in raw 3D point clouds. Experiments have been conducted on the Stanford large-scale 3D Indoor Spaces Dataset using only geometry information. Experimental results have clearly shown the superiority of the proposed network.

In recent years, point cloud data based 3D deep learning has become a popular method for three-dimensional object recognition. In this work, we introduce a multi-scale convolution neural network which takes point cloud as input for 3D object recognition. Our network structure consists of two parts which are the feature extraction structure and the feature processing part. Experiments are conducted on the ModelNet40 dataset with several state-of-the-art methods. The proposed method achieves a higher accuracy on 3D object recognition with 87.1%. Experimental results have demonstrated the superior performance of the proposed multi-scale feature learning network.

Recognizing 3D objects from point clouds in the presence of significant clutter and occlusion is a highly challenging task. In this paper, we present a coarse-to-fine 3D object recognition algorithm. During the phase of offline training, each model is represented with a set of multi-scale local surface features. During the phase of online recognition, a set of keypoints are first detected from each scene. The local surfaces around these keypoints are further encoded with multi-scale feature descriptors. These scene features are then matched against all model features to generate recognition hypotheses, which include model hypotheses and pose hypotheses. Finally, these hypotheses are verified to produce recognition results. The proposed algorithm was tested on two standard datasets, with rigorous comparisons to the state-of-the-art algorithms. Experimental results show that our algorithm was fully automatic and highly effective. It was also very robust to occlusion and clutter. It achieved the best recognition performance on all of these datasets, showing its superiority compared to existing algorithms.

2D views of objects play an important role in 3D object recognition. In this paper, we focus on 3D object recognition using the 2D projective views. The discriminativeness of each view of an object is first investigated with view saliency using 2D Zernike Moments. The proposed view saliency is then used to boost a multi-view convolutional neural network for 3D object recognition. The proposed method is compared with several state-of-the-art methods on the ModelNet dataset. Experimental results have shown that the performance of our method has been significantly improved over the existing multi-view based 3D object recognition methods.

Recognizing static hand gesture in complex backgrounds is a challenging task. This paper presents a static hand gesture recognition system using both color and depth information. Firstly, the hand region is extracted from complex background based on depth segmentation and skin-color model. The Moore-Neighbor tracing algorithm is then used to obtain hand gesture contour. The k-curvature method is used to locate fingertips and determine the number of fingers, then the angle between fingers are generated as features. The appearance-based features are integrated to the decision tree model for hand gesture recognition. Experiments have been conducted on two gesture recognition datasets. Experimental results show that the proposed method achieves a high recognition accuracy and strong robustness.

To classify thrombosis and pectinate muscle in cardiac ultrasound image sequences, a classification method based on sparse representation is proposed. This method extracts GLCM based texture features to form the sample set and compute the sparse solution with coefficients how a test sample be represented by the training set. After that, two kinds of constraints and classification strategy are added to achieve the classification. Experiment results shows that the proposed approach can achieve a classification accuracy of 91.92%, significantly higher than other popular classifiers.

With the rapid development of range image acquisition techniques, 3D computer vision has became a popular research area. It has numerous applications in various domains including robotics, biometrics, remote sensing, entertainment, civil construction, and medical treatment. Recently, a large number of algorithms have been proposed to address specific problems in the area of 3D computer vision. Meanwhile, several benchmark datasets have also been released to stimulate the research in this area. The availability of benchmark datasets plays an significant role in the process of technological progress. In this paper, we first introduce several major 3D acquisition techniques. We also present an overview on various popular topics in 3D computer vision including 3D object modeling, 3D model retrieval, 3D object recognition, 3D face recognition, RGB-D vision, and 3D remote sensing. Moreover, we present a contemporary summary of the existing benchmark datasets in 3D computer vision. This paper can therefore, serve as a handbook for those who are working in the related areas.

The directions of arrival (DOA) of interference changes rapidly with time in high-dynamic satellite navigation system, which leads to the mismatch between weights training data and weights applying data during the adaptive processing. In this case, the conventional adaptive anti-jamming algorithms are invalid since jammers may easily move out of the array pattern nulls and could not be suppressed. In this letter, a novel method of null-widen is deduced based on the Laplace distribution model of the changing interference DOA in high-dynamic environment. By taking the moving interferences as discrete interference sources obey the Laplace distribution, the extension matrix can be obtained and the average covariance matrix can be calculated from matrix product to broaden the width of nulls. The new method does not alter the noise contribution to the covariance matrix and does not require information about interference DOA. The width of nulls is easy to control. In addition, by combining with the power minimization approach, the new null-widen method does not need the DOA of desired satellite signal. The proposed method can provide correctly acquisition.