<div class="section abstract"><div class="htmlview paragraph">This paper presents a highway accident risk assessment model based on Bayesian random-parameters logit model, aiming to evaluate the effects of real-time traffic conditions crash risks freeways. By incorporating random parameters account for variations in impacts variables across different freeway segments, offers greater flexibility and adaptability compared traditional fixed-parameters models. The study utilizes flow...

Event detection plays an essential role in video content analysis. In this paper, we present our approach based on motion relativity and feature selection for event detection. First, propose a new feature, namely Expanded Relative Motion Histogram of Bag-of-Visual-Words (ERMH-BoW) to employ ERMH-BoW, by representing what aspect with (BoW), construct relative histograms between different visual words depict the objects' activities or how event. ERMH-BoW thus integrates both aspects complete...

Excessive psychological pressure, long working hours, and excessive labor intensity can make people exhausted affect people's cognition motor function. Detecting the fatigue state of athletes prevent sports injuries. This article chooses adaptive median filter method to smooth image remove noise, uses threshold light equalization adjust image's equalization. According admission rejection criteria Sequential Forward Floating Selection (SFFS) algorithm, different feature parameter combinations...

This article's goal is to investigate, using an intelligent random forest model (RFM), the logistics supply chain's network risk control method. The chain network, a vital component of modern economy, faces numerous potential risks, and uncertainties. To ensure stability sustainability chain, effective plan crucial. article examines network's management practices proposes approach based on RFM. Many data are gathered compared with other widely used models in experimental section, including...

Fingerprint classification is crucial to reduce the processing time in a large-scale database. In this paper fingerprint based on continuous orientation field and singular points proposed. The can not only filter noises point directional image, but also represent basic structural feature of more precisely. Singularities are most important reliable classification. fast algorithm made possible by simple effective combination modified Poincare index determination points.The experiment results...

Crowd density is an important factor of crowd stability. Previous estimation methods are highly dependent on the specific video scene. This paper presented a scene invariant method using Geographic Information Systems (GIS) to monitor size for large areas. The proposed mapped images GIS. Then we can estimate each camera in GIS model obtained by one camera. Test results show that be adaptively applied other cameras outdoor scenes. A real-time monitoring system areas based has been...

A rapid algorithm for eye state detection is proposed in this paper. It takes use of gray scale characteristics the and upper eyelid bending direction to distinguish open closed states. Circle similarity region as an index characterize state. And a simple method based on relation curve straight line linking two endpoints determine direction. The evaluated BioID face database, Yale part CAS-PEAL database. experimental results show that it achieves more than 85% accuracy less 5 ms runtime average.

The target detection algorithms have the problems of low accuracy and susceptibility to occlusion in existing smart cities. In response this phenomenon, paper presents an algorithm for a city combined with depth learning feature extraction. It proposes adaptive strategy is introduced optimize search windows based on traditional SSD algorithm, which according operating conditions change, strengthening enhance objective function weighted correlation fusion method, method combination appearance...

To address the issue of image denoising algorithms blurring details during process, we propose an adaptive algorithm for 3D reconstruction 2D images. This takes into account inherent visual characteristics human eyes and divides regions based on entropy value each region. The background region is subject to threshold denoising, while target undergoes processing using adversarial generative network. network effectively handles images with noise generates a model target. proposed aims enhance...

Cooperation risk evaluation of virtual logistics enterprise is a typical multi-attribute measurement problem, which relates to many attributes and the relationships among them are very complex. A fuzzy analytic network process (F-ANP) approach was presented in this paper evaluate cooperation enterprise. The F-ANP not only capable handling interdependence elements by obtaining composite weights all criteria through development super matrix but also improves human judgment on pairwise...

In cognitive relay networks, the user opportunistically accesses authorized spectrum segment of primary and simultaneously serves as data node while sharing resource user. This not only improves utilization efficiency network resources but also throughput users. However, if randomly selects node, there is no guarantee for an optimal throughput. Moreover, system power consumption may increase. order to improve optimize utility, this paper proposes a optimization algorithm based on deep...

With the rapid development of computer intelligence technology, majority scholars have a great interest in detection and tracking moving targets field video surveillance been involved its research. Moving target has also widely used military, industrial control, intelligent transportation. progress social economy, supervision traffic become more complicated. How to detect vehicles on road real time, monitor illegal vehicles, control effectively hot issue. In view complex situation various...

As intelligent vehicle networks become integral to modern transportation systems, the need for accurate and efficient multi-scene target detection in complex urban environments is increasingly critical. Addressing challenge of processing transmitting extensive image data rapidly accurately under dynamic road conditions, this paper proposes Multi-scene Target Detection Semantic Communication (MTDSC) method. MTDSC employs convolutional neural region proposal effective images, further utilizes...

Existing image augmentation methods consist of two categories: perturbation-based and generative methods. Perturbation-based apply pre-defined perturbations to augment an original image, but only locally vary the thus lacking diversity. In contrast, bring more diversity in augmented images may not preserve semantic consistency, incorrectly change essential semantics image. To balance consistency images, we propose SGID, a Semantic-guided Generative Image method with Diffusion models for...

Super-resolution methods based on sparse easily lead to over-smoothing at the edges of reconstructed image. A novel super-resolution method local and global gradient is proposed solve problem. First, it represents input low-resolution (LR) image patches with coefficients LR over-complete dictionary. Then maps high resolution (HR) dictionary reconstructs HR texture. At last, enhances main edge using natural statistics' prior information merges together By representation transformation, can...