Multimodal Large Language Models (MLLMs) are becoming integral to autonomous driving (AD) systems due to their strong vision-language reasoning capabilities. However, MLLMs are vulnerable to adversarial attacks—particularly adversarial patch attacks—which can pose serious threats in real-world scenarios. Existing patch-based attack methods are primarily designed for object detection models. Due to the more complex architectures and strong reasoning capabilities of MLLMs, these approaches perform poorly when transferred to MLLM-based systems. To address these limitations, we propose PhysPatch, a physically realizable and transferable adversarial patch framework tailored for MLLM-based AD systems. PhysPatch jointly optimizes patch location, shape, and content to enhance attack effectiveness and real-world applicability. It introduces a semantic-based mask initialization strategy for realistic placement, an SVD-based local alignment loss with patch-guided crop-resize to improve transferability, and a potential field-based mask refinement method. Extensive experiments across open-source, commercial, and reasoning-capable MLLMs demonstrate that PhysPatch significantly outperforms state-of-the-art (SOTA) methods in steering MLLM-based AD systems toward target-aligned perception and planning outputs. Moreover, PhysPatch consistently places adversarial patches in physically feasible regions of AD scenes, ensuring strong real-world applicability and deployability.

To accurately estimate full-degree-of-freedom (DOF) model parameters of the 3D point cloud acquisition device, a calibration method by using a calibrator of a simple space ball is proposed. The method can achieve full-DOF (DOF) estimation without additional hardware and step-by-step calculations. Firstly, a measurement model is established according to the rotation characteristics of the 3D point cloud acquisition device. Secondly, with the spherical constraints of the sphere, a nonlinear optimization model of the 3D point cloud acquisition device is established by adopting the bidirectional kernel meanp-power error (BiK) loss function which is robust to the measurement noise and outliers. Finally, the successful history-based differential evolution parameter adaptation (SHADE) algorithm and the Levenberg-Marquardt (LM) algorithm are combined to solve the nonlinear optimization model, and the full-DOF model parameters of the 3D point cloud acquisition device can be estimated. Experimental results demonstrates that the proposed method can accurately estimate the full-DOF model parameters of the 3D point cloud acquisition device. More encouragingly, the effect of measurement noise and outliers can be significantly suppressed.

Multi-view point cloud registration remains a significant challenge in 3D computer vision due to the sensitivity to outliers. In this paper, we propose a clustering multi-view point cloud registration algorithm based on kernel mean p-power error (KMPE) loss function. Firstly, the 3D point clouds are clustered by using the K-means algorithm, where the centroids of the clusters are regarded as the model point cloud for multi-view point cloud registration. Secondly, the model point cloud is used to estimate the rigid transformation of each point cloud sequentially. Considering that the KMPE loss function can efficiently suppress outliers, a robust point cloud rigid registration optimization model based on the KMPE loss function is established. The Levenberg-Marquardt (LM) algorithm is adopted to optimize the optimization model to obtain the optimal rigid transformation between each point cloud and the model point cloud. Finally, the clustering and KMPE-based rigid transformation estimation are iteratively and alternatively applied to all point clouds to realize multi-view point cloud registration. Experimental results demonstrate that the proposed algorithm can effectively suppress the influence of outliers on the multi-view point cloud registration of, and can achieve high accuracy of multi-view point cloud registration.

Intended to boost the simplification precision of 3D point clouds and alleviate the cavity issue caused by overemphasis on feature components, this study puts forward a 3D point cloud simplification method that combines partition strategy and information, with the simplification rate being controllable. Using the partitioning strategy, the original point cloud is divided into three regions: edge points, feature points, and non-feature points. and the three regions are simplified respectively, and the points with a large amount of information are retained by calculating and updating the mutual information amount of each point in each region. The results demonstrate that the proposed method can effectively enhance the accuracy of simplified point clouds across various types of datasets, and maintain good simplification accuracy even when the simplification rate is large.

Graph convolution neural networks (GCNNs) have shown great promise in handling point cloud due to their adeptness at extracting topological information. However, a significant limitation of current GNN approaches is the substantial time required for constructing and learning the graph structures inherent in the point cloud. To address this challenge, we present a novel graph convolution-based network designed for the efficient processing of 3D point cloud. Based on dynamic GCNN (DGCNN) and PointNet++, our network integrates graph convolution with multi-scale point and edge features for enhanced neighborhood geometric information extraction. We introduce an innovative lightweight edge convolution technique that notably reduces computational burden while optimizing processing efficiency. In addition, the network also employs a downsampling strategy within a GCNN framework and introduces a directed graph-based geometric feature extractor to robustly capture local geometries. Empirically, our model demonstrates state-of-the-art performance across three benchmark datasets, markedly improving both accuracy and processing speed in point cloud analysis. Notably, our model achieves a groundbreaking 90% overall accuracy on the ScanObjectNN dataset.

Point-based methods have made significant progress, but improving their scalability in large-scale 3D scenes is still a challenging problem. In this paper, we delve into the point-based method and develop a simpler, faster, stronger variant model, dubbed as LinNet. In particular, we first propose the disassembled set abstraction (DSA) module, which is more effective than the previous version of set abstraction. It achieves more efficient local aggregation by leveraging spatial anisotropy and channel anisotropy separately. Additionally, by mapping 3D point clouds onto 1D space-filling curves, we enable parallelization of downsampling and neighborhood queries on GPUs with linear complexity. LinNet, as a purely point-based method, outperforms most previous methods in both indoor and outdoor scenes without any extra attention, and sparse convolution but merely relying on a simple MLP. It achieves the mIoU of 73.7\%, 81.4\%, and 69.1\% on the S3DIS Area5, NuScenes, and SemanticKITTI validation benchmarks, respectively, while speeding up almost 10x times over PointNeXt. Our work further reveals both the efficacy and efficiency potential of the vanilla point-based models in large-scale representation learning. Our code will be available upon publication.

The quantum random number generator (QRNG) is theoretically capable of generating unpredictable random numbers based on the inherent uncertainty of quantum mechanics, which is of paramount importance for information security. However, the security of practical QRNG is susceptible to the influence of unknown classical noise introduced by imperfect measurement devices, leading to potential security threats. In this paper, we propose a deep learning-based prediction model for analyzing the bidirectional security of QRNG where the quantum source is contaminated by classical noise. Firstly, we train a deep learning model capable of evaluating the randomness of mixed entropy source composed of quantum source and classical source, which exhibits excellent performance and effectively avoids the limitations of Statistical tests in evaluating the randomness of the mixture of quantum and classical source results. Secondly, systematically analyzes the impact of non-ideal measurement devices on the practical security of the continuous variable QRNG, which provides an explicit basis for compensating the discrepancy between theory and experiment. Finally, we perform correlation detection on QRNG output sequences with a deep learning model and focus on both the forward and backward security of random numbers. Through bidirectional security detection, random number sequences that may be biased or manipulated can be more accurately and comprehensively evaluated, further preventing potential correlations from opening security holes for eavesdroppers.

Radiology report generation (RRG) is crucial to save the valuable time of radiologists in drafting the report, therefore increasing their work efficiency. Compared to typical methods that directly transfer image captioning technologies to RRG, our approach incorporates organ-wise priors into the report generation. Specifically, in this paper, we propose Organ-aware Diagnosis (OaD) to generate diagnostic reports containing descriptions of each physiological organ. During training, we first develop a task distillation (TD) module to extract organ-level descriptions from reports. We then introduce an organ-aware report generation module that, for one thing, provides a specific description for each organ, and for another, simulates clinical situations to provide short descriptions for normal cases. Furthermore, we design an auto-balance mask loss to ensure balanced training for normal/abnormal descriptions and various organs simultaneously. Being intuitively reasonable and practically simple, our OaD outperforms SOTA alternatives by large margins on commonly used IU-Xray and MIMIC-CXR datasets, as evidenced by a 3.4% BLEU-1 improvement on MIMIC-CXR and 2.0% BLEU-2 improvement on IU-Xray.

Different from the ground image with uniform haze, the haze in remote sensing (RS) image has the characteristics of irregular shape and uneven concentration in hazy weather. It brings a great challenge to the application of RS image data in advanced image processing tasks. A novel dehazing network for non-uniform hazy remote sensing image, named as KFA-Net, is proposed to solve the aforementioned issues. The designed asymmetric size feature cascade (ASFC), k-means pixel attention (KPA) and FFT channel attention (FCA) in KFA-Net all show excellent effects. Compared with symmetrically linked typical Unet, ASFC can more easily extract shallow features for feature reconstruction. Furthermore, different from the commonly used pixel attention that compresses feature maps directly, KPA introduces k-means clustering algorithm in machine learning into the attention mechanism, which facilitates network training to focus on the thick hazy region. Compared to typical squeeze-and-excitation block, FCA uses the low-frequency region feature of spectrogram to obtain the attention weight coefficient in the frequency domain, making network training pay more attention to the feature of image low-frequency region. Extensive comparison experiments verify that the proposed KFA-Net has the great superiority. PSNR/SSIM of KFA-Net are 31.0952% and 6.6401% higher than DCP with the highest citation in traditional dehazing methods, respectively. PSNR/SSIM of KFA-Net are 2.2049% and 0.4966% higher than the recently proposed 4KDehazing with the best performance among all comparison dehazing methods, respectively. The KFA-Net proposed in this research can greatly enhance the temporal and spatial scope of RS image application in hazy weather conditions.

Understanding what is happening in the surveillance video is important for human-machine interface in transportation systems, where temporal language grounding is one of the key tasks, targeting at localizing the desired moment in an untrimmed video with a given sentence query that is relevant to the moment. This task is challenging due to the following reasons: 1) the requirement of understanding the video contents and query semantics comprehensively, and 2) building the bridge between the cross-modal semantics. To tackle these problems, early methods first sample video clips and then match them with the sentence to find the most relevant one. To reduce the computational complexity associated with video clip sampling, recent methods directly predict the temporal boundaries of the desired moment on the fused features of the sentence and the video frames. However, all the previous methods often learn the word-level or phrase-level features of the sentence, or directly generates the global sentence representation by attention mechanisms or graph network. However, we argue that applying only word-level or phrase-level semantic information and cross-modal interactions is not enough to fully capture the correspondence between the video and the query. To this end, we proposed a novel Multi-level Query Exploration and Interaction (MQEI) model, which explores the semantics in both the word- and phrase-level and captures the multi-level interactions between the video and the query through an attention module. Extensive experiments on two public benchmark datasets ActivityNet Captions and Charades-STA demonstrate that the proposed model can outperform all the state-of-the-art methods consistently.

Nuclear instance segmentation is a challenging task due to a large number of touching and overlapping nuclei in pathological images. Existing methods cannot effectively recognize the accurate boundary owing to neglecting the relationship between pixels (e.g., direction information). In this paper, we propose a novel Centripetal Direction Net-work (CDNet) for nuclear instance segmentation. Specifically, we define centripetal direction feature as a class of adjacent directions pointing to the nuclear center to rep-resent the spatial relationship between pixels within the nucleus. These direction features are then used to construct a direction difference map to represent the similarity within instances and the differences between instances. Finally, we propose a direction-guided refinement module, which acts as a plug-and-play module to effectively integrate auxiliary tasks and aggregate the features of different branches. Experiments on MoNuSeg and CPM17 datasets show that CDNet is significantly better than the other methods and achieves the state-of-the-art performance. The code is available at https://github.com/honglianghe/CDNet.

In this paper, we propose a deep weight global registration (DWG-Reg) algorithm for poor initialization and partially overlapping point clouds registration problem. Our DWG-Reg is based on three modules: a bidirectional nearest search strategy for correspondence, a convolutional network for correspondence confidence prediction which consists of Hybird Distance Generator, optimal annealing Parameter Prediction network and a robust kernel function, a weighted optimizer algorithm for closed-form pose estimation. Experimental results show that our DWG-Reg achieves state-of-the-art performance compared to existing non-deep learning and recent deep learning methods. Our source code will open at https://github.com/BiaoBiaoLi/DWG-Reg.

In this letter, considering the inverse synthetic aperture radar (ISAR) imaging of nonuniformly rotating targets under a low signal-to-noise ratio (SNR) environment, an effective ISAR imaging algorithm based on the coherently integrated nonuniform trilinear autocorrelation function (CINTAF) is proposed. Because the definition of a nonuniform trilinear autocorrelation function (NTAF) which enables a coherent accumulation of the signal energy in both the time and lag-time domains, the proposed method has a significant antinoise performance improvement in comparison with other algorithms, while the computational complexity remains similar. The effectiveness and superiority of this new method have been demonstrated through several simulation results.

In this paper, a new sparse array structure, which further improves the degrees of freedom (DOFs) and enhanced the DOA estimation performance, for the fourth-order cumulant based direction of arrival (DOA) estimation is proposed. The new-formed array is hole-free and can achieve a large consecutive range in its fourth-order difference coarray. By analyzing its second-order sum coarray and fourth-order difference coarray, the closed form expression for the physical sensor locations and the corresponding virtual sensor configurations are derived. Compared with the existing fourth-order based sparse array structures, such as FLNA and SAFOE-NA, when the number of sensors is less than 23, the proposed sparse array can obtain longer consecutive virtual array, leading to more detected sources with a higher accuracy. Numerical simulations are performed to verify the superiorities of the proposed sparse array for fourth-order cumulant based DOA estimation.

The multiple-input multiple-output (MIMO) radar with a frequency diverse array (FDA) acting as the transmit (Tx) array, referred to as FDA-MIMO radar, is capable of providing additional degrees-of-freedom (DOFs) in range domain, and thereby offers the potential benefits in range-dependent interference mitigation and range ambiguity resolving. The existing FDA-MIMO radar literature either simply assumes that the Tx waveforms are mutually orthogonal or employs the code division multiple access (CDMA) waveforms to extract Tx DOFs. However, for real applications of the ground moving target indication (GMTI) using the space-time adaptive processing (STAP) technique, the CDMA waveforms are not preferable due to their poor ground clutter cancellation performance. To address this issue, a novel slow-time FDA-MIMO radar, which transmits slow-time phase-coded waveforms, is developed. As the slow-time FDA-MIMO radar emits highly correlated waveforms, it is expected to achieve excellent clutter cancellation performance. In addition, a new signal processing strategy is proposed, which is capable of extracting range-dependent Tx DOFs effectively. Numerical experiments are conducted to validate the effectiveness of the proposed radar framework for STAP applications.

Cooperative coevolution (CC) has shown great potential for solving large-scale optimization problems (LSOPs). However, traditional CC algorithms often waste part of the computation resource (CR) as they equally allocate CR among all subproblems. The recently developed contribution-based CC algorithms improve the traditional ones to a certain extent by adaptively allocating CR according to some heuristic rules. Different from existing works, this paper explicitly constructs a mathematical model for the CR allocation (CRA) problem in CC and proposes a novel fine-grained CRA (FCRA) strategy by fully considering both the theoretically optimal solution of the CRA model and the evolution characteristics of CC. FCRA takes a single iteration as a basic CRA unit and always selects the subproblem which is most likely to make the largest contribution to the total fitness improvement to undergo a new iteration, where the contribution of a subproblem at a new iteration is estimated according to its current contribution, current evolution status, as well as the estimation for its current contribution. We verified the efficiency of FCRA by combining it with the success-history-based adaptive differential evolution which is an excellent DE variant but has never been employed in the CC framework. Experimental results on two benchmark suites for LSOPs demonstrate that FCRA significantly outperforms existing CRA strategies and the resulting CC algorithm is highly competitive in solving LSOPs.

Estimation of distribution algorithms (EDAs) are a special class of model-based evolutionary algorithms (EAs). To improve the performance of traditional EDAs, many remedies were suggested, which mainly focused on estimating a suitable probability distribution model with superior solutions. Different from existing research ideas, this paper tries to enhance EDA by exploiting the potential value of inferior solutions, where Gaussian EDA is taken as an example. It will be shown that, after a simple repair operation, inferior solutions could be surprisingly useful in adjusting the covariance matrix of Gaussian model, then a better search direction and a more proper search scale can be obtained. Since the aim of Inferior Solution Repairing (ISR) operator is not to directly improve the quality of inferior solutions, but to make them closer to superior ones, it can be implemented in a simple way. Combining ISR and traditional Gaussian EDA, a new EDA variant named ISR-EDA is developed. Comparison with existing EDAs and some other state-of-the-art EAs on benchmark functions demonstrates that ISR-EDA is efficient and competitive.