Road Vehicle Detection and Classification from Very-High-Resolution Color Digital Orthoimagery based on Object-Oriented Method

This research proposed an object-oriented method to obtain the distribution of tree coverage in urban environment using 0.6m aerial multi-spectral images. With the support of eCognition Software, the whole tree coverage mapping process included the following steps. Firstly, selecting a set of appropriate parameters by trial and error process to obtain an optimal segmentation result for tree coverage. Then, a two-level class hierarchy was constructed combining the Nearest-Neighbor Classifier and the Fuzzy logic classifier. After classification, we created two abstract classes (tree and non-tree) and selected Error Matrix Based on Samples to perform accuracy assessment for tree coverage mapping. The result of accuracy assessment showed that the proposed method had produced 96.4% overall accuracy and 92.6% KIA. Finally, we calculated the tree coverage rate based on the statistical result of sum area of tree coverage classification.

In order to improve the speed of road vehicle and lane line detection, a road vehicle and lane line detection method with improved YOLOv4-tiny algorithm is proposed by using K-means++ clustering algorithm instead of the original K-means algorithm. Using the Mosaic data enhancement method, four images are randomly deflated and then stitched into one image to enrich the detection target background. The optimal weight values are derived by multi-scale training using the GPU through image feature extraction of the BDD10K dataset to achieve the detection of vehicle and lane line targets in the images. The results show that the improved YOLOv4-tiny algorithm achieves a detection speed of 134 FPS and an average accuracy of 77.84% mAP in highway lane line detection. After comparison, the detection speed of the improved algorithm is significantly improved, effectively improving the efficiency of highway vehicle and lane line detection.

Automatic detection of landslides using remote sensing images has been demonstrated to be an effective method in landslide inventory mapping. However, areas recognized as landslides using traditional methods may connect with each other and form a massive landslide, which is not conducive for accurate statistics on the number of individual landslides. Thus, this paper explored an object-oriented individual landslide separation method to solve this issue. The landslide areas were obtained by applying an object-oriented detection method on a SPOT-5 2.5 m multispectral image and the field investigation. A sequence of processes including the multi-scale image segmentation, spectral characteristics difference analysis, morphometric properties for landslides selection was applied. Objects detected as landslides were subsequently separated into individual landslides. Comparison of model results indicates that the proposed method can divide the landslide areas into individuals better than just merging them directly. It is also relatively quick to use this method to separate individual landslides than to use visual interpretation. Therefore, the object-oriented individual landslides separation method proposed in this paper could be a good solution for rapid risk assessment of landslide disasters.

Deep learning-based classification and detection algorithms have emerged as a powerful tool for vehicle detection in intelligent transportation systems. The limitations of the number of high-quality labeled training samples makes the single vehicle detection methods incapable of accomplishing acceptable accuracy in road vehicle detection. This paper presents detection and classification of vehicles on publicly available datasets by utilizing the YOLO-v5 architecture. This paper’s findings utilize the concept of transfer learning through fine tuning the weights of the pre-trained YOLO-v5 architecture. To employ the concept of transfer learning, extensive data sets of images and videos of the congested traffic patterns were collected by the authors. These datasets were made more comprehensive by pointing various attributes, for instance high- and low-density traffic patterns, occlusions, and different weather circumstances. All of these gathered datasets were manually annotated. Ultimately, the improved YOLO-v5 structure becomes accustomed to any difficult traffic patterns. By fine-tuning the pre-trained network through our datasets, our proposed YOLO-v5 has exceeded several other traditional vehicle detection methods in terms of detection accuracy and execution time. Detailed simulations performed on the PKU, COCO, and DAWN datasets demonstrate the effectiveness of the proposed method in various challenging situations.

Change Vector Analysis (CVA) is an important change detection method for remote sensing imagery with medium and low resolution. Traditional CVA is a pixel-based method, which is insufficient for high-resolution (HR) imagery. An object-oriented change vector analysis method (OCVA) is proposed in the paper. Image segmentation method was used to get image objects. The object histogtam was extracted as the feature of the image object at different bands. G statistic was employed to measure the distance between the object histograms at two times. Change vector of the object was built by the histogram distance at different bands. The heterogeneity of the object was measured by the magnitude of the object change vector. The threshold to detect change/no change objects was decided by Otsu method which maximizes the variance between change objects and unchanged objects. The experimental results on QuickBird images show that OCVA can make full use of the spatial information in HR images and get higher accuracy than pixel-based CVA.

Recent successes using object-oriented analysis methods to develop software systems suggest exploiting these techniques for systems engineering of large, complex systems. Unfortunately, object-oriented analysis methods are limited in their ability to model systems performance, one of the key components of the systems engineering process. To overcome these limitations, this paper proposes a systems engineering discipline that involves the integration of object-oriented models (which include information, state transition, and process models) with queueing-based performance simulation models. We discuss several techniques that promote performance modeling of hardware components as shared physical resources accessed by logical software objects. These techniques allow in-depth analysis of functionality, performance, and partitioning early in the requirements and design process which, in turn, reduces the risk, cost, and time involved in building and deploying computer-based systems.

Smart City is one of the Smart Initiatives that tries to provide solutions to parking problems that are found in big cities. The growth of the vehicle population is not matched by the growth of parking infrastructure, which raises various problems such as illegal parking, traffic jams, and driving safety. Thanks to the massive evolution of the Internet of Things, the Smart Parking idea became a breakthrough in solving the growing parking problem. In this paper, we explain the prototype development of the Smart Parking System using the Internet of Things which has three subsystems, namely embedded systems, as well as Android and web-based applications. Smart Parking System’s software was developed using the Object-Oriented Analysis and Design (OOAD) method and has been tested using black-box testing with 41 test cases that were successfully received.

Using remote sensing image for road information extraction has gradually become a significant part of intelligent highway management and maintenance. The road extraction of reasonable segmentation scale and characteristic elements are studied by applying object-oriented image information extraction method. First of all, the image is enhanced, then the appropriate segmentation and merging threshold value is selected for image segmentation, after that the road information is extracted by making rules according to the characteristic description of the road. It is proved that the object-oriented method can effectively extract the road information from remote sensing images.

Efficiency of transportation of people and goods is playing a vital role in economic growth. A key component for enabling effective planning of transportation networks is the deployment and operation of autonomous monitoring and traffic analysis tools. For that reason, such systems have been developed to register and classify road traffic usage. In this paper, we propose a novel system for road traffic monitoring and classification based on highly energy efficient wireless magnetic sensor networks. We develop novel algorithms for vehicle speed and length estimation and vehicle classification that use multiple magnetic sensors. We also demonstrate that, using such a low-cost system with simplified installation and maintenance compared to current solutions, it is possible to achieve highly accurate estimation and a high rate of positive vehicle classification.

Road blockage information extraction from a single-phase postdisaster image is difficult because roads are narrow and easily covered by vegetation. The traditional object-oriented image analysis method is restrictive, and its detection is slow. A deep learning algorithm, i.e., the convolution neural network (CNN), is applied to rapidly extract road blockage information. An algorithm for sample generation is designed to construct a typical sample library for CNN training, and an appropriate CNN structure and a complete detection process are designed to extract road blockage information. Finally, by taking the Jiuzhaigou earthquake on August 8, 2017, as an example, experimental verification is carried out. The kappa coefficient and the F1 score of the results are 77.60% and 87.95%, respectively. The extraction of road blockages can be completed with an efficiency of 14.59 km2 per hour. The requirements for disaster emergency monitoring can be met by the accuracy and efficiency of this method, which are better than those of the traditional object-oriented method.

In recent years, there has been a demand for Web applications with complex functions. In addition, most web applications efficiently manage data based on databases. While the key and critical dimension of developing these Web applications is analysis and design, most object-oriented analysis and design methods do not have a consistent view of the database. In addition, Java Enterprise Edition (EE) -based technologies are used in Web application implementations, but they do not provide any correlation with the database. On the other hand, as users' demands for security increase, security becomes more important. To this end, Java EE and database systems provide security solutions. However, it does not provide any correlation with object-oriented analysis and design methodology. As a result, it is difficult to develop secure web applications in a consistent way from analysis to implementation. In this paper, we propose a consistent software development methodology from analysis to implementation of secure web applications. The proposed software development methodology for web application development uses UMLsec, a security-emphasized modeling language, and object-relational (O-R) mapping for relational database design. It also uses Java servlets and SQL to implement analysis and design results based on role-based access control (RBAC). The software development methodology for the secure web application proposed in this paper has been applied to the development of the online banking system, from the design stage of the user's requirements analysis to the implementation of the web application.

Vehicle-activated signs (VAS) are speed-warning signs activated by radar when the driver speed exceeds a pre-set threshold, i.e. the trigger speed. The trigger speed is often set relative to the speed limit and is displayed for all types of vehicles. It is our opinion that having a static setting for the trigger speed may be inappropriate, given that traffic and road conditions are dynamic in nature. Further, different vehicle classes (mainly cars and trucks) behave differently, so a uniform trigger speed of such signs may be inappropriate to warn different types of vehicles. The current study aims to investigate an automatic VAS, i.e. one that could warn vehicle users with an appropriate trigger speed by taking into account vehicle types and road conditions. We therefore investigated different vehicle classes, their speeds, and the time of day to be able to conclude whether different trigger speeds of VAS are essential or not. The current study is entirely data driven; data are initially presented to a self-organising map (SOM) to be able to partition the data into different clusters, i.e. vehicle classes. Speed, time of day, and length of vehicle were supplied as inputs to the SOM. Further, the 85th percentile speed for the next hour is predicted using appropriate prediction models. Adaptive neuro-fuzzy inference systems and random forest (RF) were chosen for speed prediction; the mean speed, traffic flow, and standard deviation of vehicle speeds were supplied as inputs for the prediction models. The results achieved in this work show that RF is a reliable model in terms of accuracy and efficiency, and can be used in finding appropriate trigger speeds for an automatic VAS.

Object-oriented structural and geometric analysis methods for truss structures are proposed. The methods are based on the iterative relaxation of nodal unbalanced forces by moving the nodes. The entities appeared in truss analysis are modeled as classes, and their class hierarchy is established. The knowledge about a truss node, a truss member, a material, and a sectional shape of the member are extracted and constructed as a knowledge base in an object-oriented manner. The object-oriented structural analysis is found to be valid for material and geometric nonlinearities, and the object-oriented geometric analysis where a kinematic problem is converted into a problem for a force equilibrium at the nodes is also found to be effective for determining the geometry of variable geometry trusses. The whole system is written in an object-oriented language, Smalltalk, and several results on structural and geometric analysis are presented. The speed of the computation is discussed along with the size of the structures and the over-relaxation factor. The extension of the knowledge base is also discussed.

Detection of road vehicles at a railway level crossing by means of optical fiber distributed sensor

Vehicle detection in driving simulation using extreme learning machine